ADR 0313 makes the span from the move to fixing to delivered the operator's
target; the delivery budget stays a measure inside it, and its leaf figure is
three minutes since ADR 0304.
The knowledge base's whole-set cited rule (to-be 55, ADR 0313's own review)
refuses to accept a record nothing cites. Design 47 already builds on
ADR 0282's delivery budget, which ADR 0313 extends with the operator's
wider fix-time target, so it is the record's home.
The operator decided on 2026-10-12, on issue 504, that a rule of the
record set that fails is a fault of the records, not the module failing:
health says only whether the module works (issue 508's principle), so a
finding must not hold a walk at the module's first-node gate.
The knowledge base's cited rule failed for ADR 0303, 0306-0312, 0314 and 0315: accepted,
and nothing cited them. Each design that owes the decision now names it in its decisions
and says it in prose, so the records can name these designs as their homes.
to-be 45 section 2's field table put the count of observations, the last observed
time and the evidence in the conditions bucket, which made every look a write and
broke ADR 0289's rule that a bucket is written only on a change.
A token a media server got from its vendor could only be typed in per consumer; the vault now takes it
from the provider, sealed to one call, and keeps it as the provision's one credential.
ADR 0302 made Media a domain: content, kind, media release, library,
media file, media-importer, download client, engine and grab now have
one meaning each. Release already meant two things, so the Media sense
is qualified in the homonyms.
mesh/repo-check pass: THE CHANGE ALTERS ITS OWN CHECK (merge-check.sh): main's version judged it; the change's judges the pull requests after it merges; it…
mesh/repo-check pass: THE CHANGE ALTERS ITS OWN CHECK (merge-check.sh): main's version judged it; the change's judges the pull requests after it merges; it…
ADR 0290 rule 9: once the import is checked record by record, hq keeps no copy.
The records and efforts are removed; every link or path to them reads "ADR NNNN" or
"research NNN"; a design's decisions: lists "ADR NNNN". 02-DECISIONS keeps a pointer
README with the reading order's six topics, which the knowledge base reads through the
records module; 01-RESEARCH keeps a pointer README. The checks drop the rules that
judged records (the knowledge base holds each at propose; checks/README.md says where)
and refuse a record written here again. Playbooks 01 and 02 and the skills write and
read records with the knowledge base's verbs.
The operator could not tell from a notification whether its cause was
known, tracked and being fixed. An issue tracks a condition by its key;
the tracker answers which issues track which conditions, and the router
says it in every message about the condition. Amends to-be 53 and 45,
playbook 03 and the glossary, and points ADR 0253 at the new line.
Nodes, assignments, walks, builds and settings could be seen only by polling verbs, and a setting
changed with no event. Each becomes a record mirror: a bucket written from the store after the commit
and only when the value differs, one event per write, compared with the store on every self-check run
(D17, mirror-drifted, healer H6). The write-only-on-a-change rule from issue 326 now holds for every
bucket; the conditions bucket breaks it and issue 420 tracks that. The effort stays active for G1, G3
and the views.
Graduates research 039. The tool runner (and the controller for its own verbs) asks the operator
through the existing operator-channel asks and runs the exact call once on the warrant, reusing the
act digest and warrant check of ADR 0259 and the shown-in-shape form of ADRs 0274 and 0277, so
risky agent calls stop resting on the agent's restraint. Amends to-be 46 and hands it off.
The review found questions no record decides: 012's minimum-viable-node
half and 027's packaging, secrets, sizing and leftovers questions, so both
stay active. Where the record lives (006) is issue 414, and 037's two
other leftovers are issues 415 and 416; to-be 06 now points at issue 414.
Ten graduated and two abandoned efforts still said active, so the record
read as open questions that later ADRs and designs had answered. Each now
names in became: what decided or overtook it. To-be 45's undecided item
moves from research 017 to issue 409, and research 037's shell follow-up
is issue 410.
The live module check refused mesh-catalog#190's smb-users setting (ADR 0262);
ADR 0283 decision 7 gets a dated insight, to-be 52 and the 0263 note follow.
Two agents could start on one issue without seeing each other, and an issue
could be resolved without anyone having said they started its fix. Adds the
status fixing, the issue's worker (by, who@node) with take-over and a stale
bound, and resolved only from fixing; settles what to-be 53 left out as
claims, under the new glossary word worker. Amends to-be 53, playbook 03 and
the hq-diagnose and hq-status skills, with pointer notes on ADR 0275 and 0281.
A module whose only unhealthy part waited for the operator's secret failed
its gate, held every later walk and told the operator there was nothing to
do. The record adds a waits answer and a waiting state, checked by the
controller, read by the gate as waits for a person, and said as a
needs-operator condition naming the act; and secret families so each SMB
source's password is one hidden line at the desk prompt.
The operator accepted them on 2026-10-10. The runtime seat and the default
half of declared settings are live; the rest is tracked by issue 398. To-be 42
cited them only as waiting, so it now names them.
From the independent review: a core delivery could wait on a leaf judgement past
its ten minutes, the window figure's split was not stated beside it, and the
one-repository rule was credited to the wrong issue.
The operator asked that a bug fix land on every node in under five minutes. 51
walks measured from the controller's own records put a leaf walk at a median of
6.0 minutes (7 of 20 within five) and say where the time goes: waiting behind
another walk, the first-node judgement, the merge window. The record weighs each
lever by what it would have saved, keeps three readings for core modules (ten
minutes, the measured worst being 8.8), and proposes the measure and condition
that check it. To-be 47 gains the section, the owners and five slices; the
glossary gains delivery time, delivery budget, core and leaf module.
The record is read-only and hand-written; numbers collide and private knowledge
has no home. Framed as questions, weighing git as the store against a store
with hq as its public export.
Every controller change walks the core's tier and holds up every other walk
(issue 382), while most of the week's change sat in one package. Framed as
questions: which domain changes most and needs the core least.
The operator approved measuring, from the delivery's own events, whether each fix
an issue names reached the machines; delivered is not resolved, and a merged fix
that does not arrive must be said rather than found.
The import of slice 3 (ADR 0275) was checked against every folder: no report changed since the
commit it read, and the tracker holds each report's status, owner, fix, design amendment, replay
and closing day. The keys are removed from all 361 reports, cycle.py refuses them in any report,
and the playbooks, skills and READMEs that still pointed at them name the tracker instead.
The operator asked for severity and what an issue needs to be estimated by the
judgement that already reads the backlog, and for dependencies between issues,
an issue being workable only once what it waits for is on the machines.
Review found the two questions together exceed the consumer's ack wait, which the
tool runner did not extend; the record says so and names the fix it waits on.
The operator asked for duplicates to be consolidated, and better, when an issue is
opened rather than on a schedule. A judgement needs a model, and the mesh had no
channel for one; the agent module answers one question with no tools, under the
node's licence, and the tracker acts only on a clear duplicate, closing the newer
issue and keeping its report on the older one.
The operator does not want to run manual commands on the control node. Issue 339 made every trusted
setting the terminal's alone, and a module's own secret the terminal's or the give verb's: each a line
an agent could only hand the operator to type. ADR 0274's pattern is extended: an agent proposes, the
operator approves on a channel that proves who answered, the controller sets the layer on the warrant
bound to the exact values shown. A secret is asked for by a verb and typed at the desk, bounded, the
prompt naming who asked. The designs 44, 52, 46 and 13, issue 339, ADR 0259 and 0266 and the glossary
are amended with dated notes.
Issue numbers collided while each author read main and the open pull
requests; since 2026-10-10 the tracker (ADR 0275) hands them out. The
playbook and the hq-new-issue skill now open every issue there, keep a
folder only for a design or governance issue, and say where its status
lives until slice 3. Research 042's finding that the SDK had no
compare-and-set is corrected, and to-be 53 names the module's code.
The operator named the module on 2026-10-10: the mesh's own parts carry the
mesh- prefix, and the seat is the role. A dated insight in ADR 0275 says so,
and that compare-and-set state was already in the runtime (ADR 0259): slice 1
added the SDK's Increment. To-be 53 and the glossary say the same, and the
design is handed to mesh-catalog and mesh-sdk (in progress).
Two catalogue merges 18 s apart opened two walks; the second superseded the
first and nothing gave it its word for 58 minutes (issue 362). The operator
decided that merges are collected in a rolling merge window (90 s, at most
10 min) and each batch is walked once, one commit per repository (the later
one of a repository, which contains the earlier), one walk at a time; the
walk-to-delivery bookkeeping is repaired too and every wait has a condition.
Amends to-be 47 and 10, adds merge window and batch to the glossary, and
corrects the issue's account of the S16 condition.
Numbers taken by reading a repository many branches write at once collide;
one counter in a module of the mesh's own, advanced by compare-and-set,
ends that at the source. The operator delegated the five open questions
on 2026-10-10; each is decided as the research recommended. The glossary
gains the issue tracker, retires "ticketing system" and lists ticket as a
homonym, since ADR 0258 already uses it for an answer's one-time token.
The operator asked for mesh tools to add shared instructions and context. Issue 340 made that text the
terminal's until a surface of the operator's own exists; the operator channel is that surface. Proposed,
for the operator to accept.
Opens the effort the operator asked for on 2026-10-10: an issues module on the bus that hands out
numbers atomically, so agents stop colliding on them, and takes in the 357 issue folders. Findings
cover the first slice, where the truth lives, the import, conditions, views, and hits as a model.
The setuid search did not finish within its bound on the control-node, so no
approval proven there is honoured; a confined start may answer the question
ADR 0259 asks directly, at a cost that does not grow with the disks.
The desk's actions sit on dunst's middle click, which a touchpad cannot give, and the context menu
names a command dunst's PATH does not hold. Located in the dunst module; dunst's own code settles
what a tap with do_action does.
The confirmation review of issue 356 found that the by-path chmod and chown of an ordinary mesh directory outside a home follows any link in its path. Issue 356 closed it for a handed-over directory, on every apply; the general gap is older and wider, and is recorded here rather than fixed in passing.
The review of 2026-10-09 proved a responder can reach the hand-over's subject through a reply, and that the apply followed a link the hand-over had refused. The record no longer says only the controller may publish there: it says the ask is signed with the mesh's key and verified by the engine, and that a handed-over directory is changed through a descriptor that followed no link. To-be 45's writers table gains the row for a hand-over asked of a machine.
cycle.py skipped a rejected option only when "Options rejected" was its nearest heading; a record that
lists its options under ### subheadings of "## Considered Options" had them held as deferrals.
A review found the check too lenient: an issue cited as context beside a deferral passed it, and one
insight excused every deferral in a record. The tracking issue is now named as tracking it, an insight
covers only its section or what it quotes, and the phrases include "not decided here", "postponed" and
"future work". ADR 0273's four undecided items get issue 357.
ADR 0244 left the code renames of the node-engine and the tool runner for later with no issue, owner or
date, and two days on the operator believed them live. Open the issue with the inventory, the risk and a
staged plan; add the rule to playbooks 02 and 04, as a dated insight on ADR 0244, and as a cycle.py check
that found the same gap in ADR 0270 (issue 355).
Review of hq #254: a hand edit is written back on the look that finds it, and a key that
decides what runs is never taken over; the guard and the condition cover settings.local.json,
plugins and plugin installs; after a node's first adoption an unrendered file is moved to the
copies; autoMode is a permission rule; the named leftovers, the owned and excluded entries,
pending requests with the exact nox line, and the agent's existing corrections as first lessons.
The engine on every node's PATH was the one a person copied at installation, a build behind every delivery; and the way out of issue 339 was a terminal act at the machine, which ADR 0272 and the operator's rule of the same day make impossible to type. The record, located in the engine, the controller and the catalogue's check, with the fixes' pull requests; ADR 0272 gains the rule as a dated insight: the operator types one command, nox, and every text addressed to the operator names a nox line. words.py now refuses a daemon's line, with --list lines for the catalogue's copy; the lifecycle and the operator's-machine designs and issue 343's quotations are reworded.
Research 040 measured the homes on four nodes: a skill retired a week ago on all four, the
operator's rules on one, nothing raised. The module now owns every configuration file under
the agent's home, excludes the program's working state by name, and raises agent-home-drift
on a hand edit; a fourth scope carries the operator's personal rules to every node; an agent
proposes a lesson when corrected and the operator accepts it at the terminal or on a warrant.
To-be 36 §10 and to-be 40 WP7–WP10 say what is built; ADR 0216 and 0243 gain notes.
The 2026-10-09 16:19-16:36 sequence on the control node: a release's gate
held the machine's report against a newer send another plan made, failed
three healthy builds and put them back to a controller that could not
read the store; that controller then passed the newer plan's gate from
its own health. Measured and guessed apart; the fix in review.
The catalogue's #125 stayed held 28 hours after ADR 0270 went live. The record covers a held merge when a
later delivered walk sent its module everywhere, as the machine steps say; a build held back for a person
and sent on the operator's word travels by a backlog walk, which gives one step per machine and whose end
mesh-delivery read for nothing. Report, diagnosis located in mesh-delivery, the fix in mesh-catalog #163,
and a progressive insight on ADR 0270's one-machine consequence, which this widens.
Plans of a branch superseded each other by when they were made; the
catch-up's late plan of an older merge took over a newer one, once a
security fix to the forge's module. Recorded with the fix in review.
Records the 2026-10-09 outage on the control node: what was measured
(the unit ran, the private address refused; a restarted node-engine's
undecided statement cleared the network condition and its reopening
dated the fault after the send), what the reproduction showed of
bind-dynamic, what is still guessed, and the fixes in review.
The confirmation review: hq requires only mesh/repo-check, so a watcher
that looked for mesh/merge-gate alone would never have said this
repository's own unchecked pull request. And the delivery domain's own
condition, a delivery held past its bound, says it rather than a provider
standing.
An ordinary line meets the generic verb's reading forms and never a
terminal-only command (node account and node agent-account by ADR 0266,
token, operator, a secret's export); a terminal line drops ADR 0266's
served mark, so it reads as the terminal. How it is checked says the
withholding tests search base64 too and fail when what they guard goes,
and the control-node refusal is tested through the acts.
- §3: the node-engine vouches and decides nothing; deadline first, other
namespaces refused, each account bounded and answered busy; the subject
is the machine's own in the writers table.
- §4: phase 1 refuses the agent account (phase 2 answers it); an ordinary
settings line takes the settings verb's path; the controller's module is
assigned at the terminal only, so the terminal does not follow assign;
'no wider than today' holds only while its two measured facts hold.
- §5: every line is a call, followed to its answer; 'nothing ran' only of
a line nobody took, otherwise 'not known' with the call; the answer is
its asker's alone.
- §7: after the four merge, the operator assigns mesh-cli to every node.
- How it is checked: each claim with its test or measurement, the
MESH_VERB refusal in mesh-cli named a courtesy, and the sudo and
agent-account facts read on the control-node today.
The forge's announcer reads only repositories whose update time moved; a
pull request opened after its push does not move it. Located in the forge
module, with the fix (mesh-catalog) and how it is checked.
Found in the review of the picom settings: a backend of xrender"; write-pid-path = "… adds an option of
the caller's. The two files marked trusted false on main are judged (user-dirs.dirs open, dunstrc not),
and typed settings validated at the verb and quoted at render are proposed; until then nothing new goes
trusted false.
The record said a merge needs a person; the forge requires no approval and
the merge tool merges once the checks pass, so nothing enforces the
operator's yes. What still remains now says so, and that the restart fix
counts from when the controller first saw the search pending. Item 10 and
routes A and B describe the follow-ups of the review of 2026-10-09.
The operator reports picom leaving a chat window stale after a workspace switch before, never recorded.
The diagnosis now names the graphics stack, puts picom first among the candidates, and makes the first
step at the next empty reminder turning compositing off, which loses nothing; a Betterbird restart comes
after, refused while a message is being written unless the operator says yes.
Records the symptom, what four reminders raised from a throw-away profile showed (drawn every time, at
depth 32), what that rules out, the candidates left, and what to measure at the next real reminder.
Status diagnosing: the cause is not known.
The records send the operator to "the controller's terminal", and on the
control-node `mesh-controller` is not a command: the controller is a mesh
process with its own account and environment, and nothing puts anything on
PATH. The record proposes mesh-cli, a module on every node whose program
asks the node-engine over a local socket; the node-engine vouches for the
account from the kernel and asks the controller on its node's own subject;
the controller runs the line as the terminal only for the operator's account
where no agent can be that account. The glossary gains mesh-cli, nox (its
zsh alias) and the controller's terminal.
Item 10 said two minutes; the engine's bound is 15, with a two-minute limit
inside it still to settle. What still remains now names a build from another
repository registering a module (mesh-controller 174), the forge's tools
changing its protection (mesh-catalog 153), and the urgent condition every
engine restart raises (mesh-controller 175).
The routes found around the agent account since the record was written are
closed on main (the store's reader, places and served keys, what every session
obeys, execute withheld on the control node); the record names them, drops the
store gap it said kept the boundary from holding, and says plainly what still
remains, first the login shell's execute on the other nodes. To-be 36 §9 gains
the prerequisites and the order the parts go out in.
The re-review closed the command verb to reads, moved enrolment to the
controller's terminal, widened what the node-engine judges, and found the
store's reader able to read every secret; say each, and that the boundary is
not in force until the store is fixed.
The review found ways round the record: a verb that set the accounts, root
following a link below a home, a tool reading any path as the operator, a
restore written beside the agent's files, a stale verdict passing. Record each
bound, the exact ways judged and not judged, the guard on every tool and its
record's limits, and a rollout order the guard does not block.
The operator chose this remedy for ADR 0259 section 8: an agent there could
become root through the operator account's sudo and docker group, and read the
tool runner's credential. Record the node fact, the node-engine's judgement
the self-check reads, the guard, and how a session starts; issue 333 holds
the measured symptom and to-be 36 the design.
The confirmation review found the home scope wrote the operator's own
~/.claude on nodes where the agent is the operator, and that text put
straight on the bus at the mesh or node scope still renders; record both,
the fix for the first, and what closes the second.
Shared agent text is the terminal's until a surface of the operator's own
updates it, the wider controller rule stands, and a password is not a
setting: 342 names every module that takes one through a setting.
The operator accepted both on 2026-10-09. Design 47's state table gains covered and what ends a held
delivery by itself, and its group the wait for a composed check and the lift; design 45's signals table
gains S17 and its planned bus step says the wait is told at once. The glossary's delivery entry names its
six final states.
Decision 1 named S6 for a failed new-module build; S6 is a lost build ask. Such a build is said by one
line in the controller's log until its module is assigned. A module on one machine never gets a 'rest'
step, so covered never reaches it. The forge module's tests run in the catalogue's TypeScript check, #151.
Node-engine #66 passed both required statuses after #64/#65 had merged,
and its merge did not compile: each side added Applied.Declared. The
check clones the head alone and nothing marks a verdict stale when the
trunk moves; ADR 0237/0238 say the head, so the fix needs a decision.
Merge order on a protected trunk is kept only as the first filter; the forge's compare, through a new
read-only tool, decides whether a later merge contains a held one. A merge held without a passing check is
covered only by a later delivery whose own check passed. Covered says a later merge that contains it was
delivered, not that the change runs: a revert in between stays possible. A delivery that only adds a module
is delivered whatever its build does, and covered wins over go when a person's walk fails between two
readings. The glossary gains covered beside superseded and delivered.
ADR 0271 names the bus upgrade as the mesh call and says S17 clears once the step sent the new bus.
The review found that a mergeable file takes any key, not only those it
names, and that the text a session reads stays registrable through a verb;
say both, and the follow-ups the module took.
Issues 319 and 316: fifteen held deliveries asked a person a question with no right answer, and stop,
the only way out, records a delivered change as an error. ADR 0270 delivers a merge that moves nothing at
once, adds the final state covered for a merge a later delivered walk of the same trunk carried, and lifts
a group hold once its composed check passed.
Issue 336: a send held for the bus's planned step raised nothing for 28 minutes. ADR 0271 adds row S17,
said at the first refusal for the operator with the verb, keeps S3 from calling it lateness, and has the
delivery plan say before the merge that a person's bus upgrade follows.
Each issue's diagnosis says what the fix does and how to verify it once delivered.
Record the hole the review traced, why the module cannot tell who called,
which kinds run or are obeyed in a session and which stay registrable,
the fix in the controller and the module, what is registered today, and
that it must land before ADR 0266's agent account is switched on.
Each fix's pull requests are merged and delivered; each issue now says
what was read live on 2026-10-09 to show it, so the record stops
listing fixed faults as located.
The operator accepted ADR 0269 on 2026-10-09. ADR 0118 gets the dated note the README asks for where a
later record narrows its mechanism. The node-engine's to-be design names the decision and says what it
changes, which the cycle check requires of an accepted record.
Third review: a person's stopping a unit the mesh also held stopped leaves no trace, so it is started
again on removal; the home server's restored weekly timer would start a scrub of about 46 hours at once.
A masked or since-changed unit is left alone, and a failed restore waits.
Unassigning the zfs module would leave the home server's pool with no scrub schedule: ADR 0118 never
starts a unit on the way out, so the distribution's timer the module held stopped would stay stopped.
The host now records what it declared beside what it found, and gives back only what the mesh's own
declaration stopped.
A new module counts as something to move, so its delivery is held; a later walk
reaches only the delivery whose merge is its commit. Four covered deliveries were
stopped before this issue was found, and their merged pull requests now say error.
The operator's decision of 2026-10-09: execute runs any command as an
account that can become root without a person, so the control-node stops
offering it now, and every node marks it as needing approval once the
mark exists. To-be 41 gains WP7 for the seat and the shell module.
Located in the controller (places and accesses taken through every verb, no line-break check) and the node-engine (a found directory chowned as root, through a link too, and any path mounted). ADR 0266's boundary depends on the fix.
A merge to a shared repository moved every module that read it, in a gated
tier each (issue 338). The operator's rule maps a change onto the artifacts
whose build source it touches and adds only modules with a build dependency
on a changed artifact; sharing a repository is no dependency. Supersedes in
part ADR 0162 decision 1 (the packages kind) and ADR 0238 decision 3 (whole
repository and context reads); adds build source to the glossary.
Adds what the fix cannot see (definer functions and views; the name check
as a backstop) and, as a follow-up, that the reader's read-only session is a
default a statement can turn off, so the missing write grants are the guard.
A controller merge walks three gated tiers because the route proxy and the
build seat are planned by repository, not by the paths their builds read,
and their programs carry the commit. Located; fix directions recorded.
Opened and diagnosed: pg_read_all_data on the login the store seat's query
runs as. Says what the store can and cannot tell about whether the key was
read, why it should be rotated once the fix is taken, and that no rotation
mechanism exists yet. The fix is mesh-catalog PR 143.
For 28 minutes on 2026-10-08 every send to the control-node waited for a bus upgrade only a person can start, and nothing but a walk's note said so. Located in the controller: the refusal is kept as the walk's note, S3 fires only past a 30-minute bound with generic words, and probe DB reads only a running step.
The record said its data is still copied; the shipped follow-up backs it up as this controller reads
it and lists an unplaceable line rather than failing the machine.
The record claimed a left-out module kept what the machine holds for it and that an entry's unknown key
was read like a top-level one; neither is true of the merged controller.
Dropping an unknown key would run a module without something its manifest says. Also name the check by
what a key is about, narrow the listing to one machine, and say how to roll back below this record.
The operator asked for one place that lists every module's preferences and each machine's value, and
for a list of the values the catalogues hard-code that should become preferences.
A stored manifest read strictly would fail the whole catalogue after a controller rollback, and a node
called default would lose its layer. Also scope the default, retire tunable, and list the new checks.
A ${setting:} nothing set failed composition, and settings refused a key no file asked
for yet, so no order of changes could add a setting to a running module. A manifest may
now default a preference; the operator's own value still has none.
A leftover user unit stayed raised on the workstation: no seat verb showed
the dangling enable link that keeps asking for it, or reset its failed
record, so the only step left was a shell. Lands per ADR 0246.
The review of mesh-controller#150 found the controller now makes an
assignment after the person's act, which no record decided. Decide it,
cite it from the glossary and to-be 45, and correct issue 325's account
of what is settled where and what is sent.
The bus's own numbers were unexplained and raised by nothing. Measured again
read-only: mesh-delivery rewrites unchanged deliveries (326), every controller
verb dials its own connection (327), API errors have no reader (328), event
consumers outlive their declaration (329), and design 25's dead-letter stream
was never built (330). The two slow consumers join issue 270, whose missing
system account they share.
The SAS sentence now holds for modes with spaces, a hung read is unhealthy in two to three minutes, not
one, every USB drive is told sat, and the limits of a stuck read are recorded.
After the second review: smartd is dropped as a second reader that could wake or keep awake the drives,
a stuck read or a drive not looked at is unhealthy, a drive asleep for a day is HIGH and never unhealthy,
and a board chip's maximum says nothing of a fan. The never-woken and hourly claims now say exactly what
holds.
Review found that ADR 0253 let any process at the operator's session release,
stop or start a delivery in the operator's name, and that its tokens, fallback
words and quiet warnings failed on the paths that go wrong.
Rule 4 now matches the module after review: the NVMe temperature bit, old-age attributes, a fan's minimum
or alarm and a board chip's limits raise nothing; a stale watch does. The standby caveat is recorded: no
mesh tool sees a drive's spin-down timer, so spinning drives are read hourly.
The store's tools counted 23.8 GiB nothing records: copied bases the sweep could never
let go of, platforms a let-go index left behind, and images from before the mesh. Package
retention read no lockfile, so it kept versions by age alone.
The operator read the reworded popup and could not tell whether to act, and was
sent to open a session to act at all. The record now decides the verdict, the
quiet rule and the path by which a notification is answered.
A machine's temperatures, fans and drive health were read by the laptop's vendor module only; the
machines' 20 drives were watched by nothing. The sensors module reads them on every machine, holds no
seat until a caller must reach it by role, and says a hardware fault through ADR 0240's health until the
machine judges its own hardware. To-be 48 gains §11 and phase G.
The operator chose a generic boundary for the bar: a block is said in terms no bar
owns and rendered by the seat's holder, placed only where the machine reports the
hardware, and generic hardware comes from the generic seat's module, not a vendor's.
To-be 42 records it; ADR 0208 points to it.
The operator could not read the mesh's notifications: they were the summary meant
for an agent, with ids, commits, keys and verbs, in UTC. The condition now carries
a headline, an explanation and a resolved line beside its summary.
Issue 247 left adding the operator's account to a daemon's group a sudo step by hand. Decide that
a module declares it with the user resource's groups, that the mesh gives back only what it added,
and that the node-engine says relogin needed; amend to-be 41 with WP6 and resolve the issue.
An answer over the bus's max_payload was refused inside the controller and
every caller timed out, while calls said answered. Diagnosed and proved (R314),
and told apart from issue 200, whose permission refusal issue 265 already fixed.
Liveness judged only long-running resources, so a module whose daemon
is a package's unit failed at every login while its machine read
healthy, and a degraded service manager was said by nothing.
The build seat ran the merge-check.sh in the change's head, so a branch older
than the script, or one that gutted it, merged untested; and the media
catalogue did not require mesh/repo-check at all. Located, fixed on branches,
replayed as R310; the protection change awaits the operator.
A group of the controller, the node-engine and the lab, each ready, was
refused as a cycle: built by put the controller's member first, engine
before controller the node-engine's. ADR 0249 ranks the rules (declared,
then built by and version skew, then engine before controller) and
supersedes ADR 0239's sentence that made a declared order against an
inferred one a cycle; design 47 says the precedence. Resolved by
mesh-controller #134 and mesh-catalog #119, replay R309 (mesh-lab #64).
Store-bound packages ran five to seven times slower there while the rest
ran alike: the throwaway store flushes to a disk the mesh's own data keeps
busy. And a holder recreated mid-check left its store and bus running.
The branch building ADR 0169 carried a fix making a node's declaration
consumer as it enrols. The controller already asserts every node's
consumer before each send, and the self-check's healer asserts a missing
one again, so the fix is dropped rather than merged and the record says
why. The installer now places the bus's accounts at genesis.
Two incidents in two days had one shape: a change took effect that nobody saw first (issues 241
and 304). Three guards where the change is made — a settings layer read before it is replaced, a
push that says what it will change, a running module's data move acknowledged — each silent when
nothing is at stake. Accepted by the operator on 2026-10-08.
To-be 44 is in progress in mesh-controller (feat/no-change-takes-effect-unseen), migration 0078,
fitted onto what the controller does since: a whole-mesh push that says so and leaves a machine
waiting for a gate, a person's push that says what it recreates (ADR 0242), a named push's cascade.
The decision index is generated (ADR 0248), so no line is added to its README.
303: a media server's previews were reached through a link its container never mounted — fixed by
mounting them. 304: settings set replaces the whole layer with nothing to read it first and no
history. 240 is fixed by mesh-controller#47; 242 has backups running and records what the rollout
taught. (First opened as 245 and 246; main has since taken both numbers, and 302 was the last.)
292 and 298 resolve with replays now registered in mesh-lab. The other fixed
core issues say why no replay can be laid over the commit before their fix. The
rest are re-checked against main.
The operator asked on 2026-10-08 for the deprecated issues to be gone. In hq an issue
is closed, never deleted, so each one is resolved with its fix, closed wontfix with a
reason, or re-checked against main and the live mesh.
301: S15 counted the operator's pushes of recorded builds, the word ADR 0242
asks for, as repairs and wanted a healer for three causes. 302: the lab's
replays register was never compiled before a merge, because its check ran in
the TypeScript toolchain. To-be 45 §7 and §9 say how each is now decided and
checked.
Research 033 measured the laptop's VPN client and the mesh overwriting each
other's resolver file in nine of nine sessions, each time cutting off one
kind of name. The operator chose a resolver module on a node seat of its
own, installed only where something requires split-dns, with the VPN
client's module carrying the adapter. Graduates research 033 into to-be 50,
and amends connectivity §2, to-be 48 §10, the seats and the glossary.
The controller that merged (mesh-controller #116) dropped the staged seat and
marked both verbs optional, stored in the seat's row and read back optional.
words.py matched a retired word only when nothing followed it, so the
plural of every retired word passed. The pattern now takes s or es on the
last word; 14 uses in 9 documents are reworded, two of them 'the hosts'
for the node-engines.
Design 33 said a seat's verbs change additively, and section 3 made each
one a condition of holding, so a verb added in the controller and served
from the catalogue could land in neither order. It happened three times on
2026-10-07. Section 7 now says the three steps and how each is checked.
The written decision index is the one file every decision PR edits, at the
end of its topic, so on 2026-10-07 each merge forced the others to rebase
and be checked again. The diagnosis weighs four fixes and recommends
generating the index on read.
The agent reached machines over ssh because search never found the verbs by the commands it knew.
Records the three rules (generated instead-of table, search by replaced command, the shell guard
with the operator's recorded override) and how each is checked; to-be 36 names it.
The plan delivering the first health declarations recreated the database and
the document store (record) and nine of mail's containers at once. Record
held only against a person's push to another machine; it now holds against
every send but a person's, and a send says what it recreates.
plans counted a plan's time from its last save, and every advance saved it, so a queued build read
as seconds old and was never late; located in the controller, fix in mesh-controller PR #119.
ADR 0244 sorts the mesh's concepts into ten domains (ADR 0006's contexts
carry over as domains), keeps machine and node as distinct words, and
makes the glossary the authority with every retired word on its
replacement's Not: line. To-be 49 draws the domains; the glossary is
reorganised by them, its two contradictions removed and the missing
words added. words.py now fails on a retired word in running prose and
on a word defined twice, so the rule is enforced rather than believed;
the 63 documents it failed on are reworded here, and research 034 is
kept as the record of the words it studied.
The operator asked for one language per domain because the same concept
keeps arriving under different words. Inventory the words in use, test a
domain split, list the clashes with evidence, and describe how the
glossary rule would be checked.
The laptop's VPN client replaces the mesh's resolver file on every connect and
the node-engine writes it back minutes later, so either the mesh's names or the
corporate ones fail for whole sessions. Measure how the client sets DNS and
weigh how names can be routed by domain without breaking ADR 0223's one-answer
rule.
The mosquitto module passed its broker's admin password to every docker
exec as an argument, which the container runtime records in its events.
Records the cause, the other modules found doing the same, the fix
proposed in mesh-catalog #100, and where the leaked value went.
The gate judged a module by what the mesh sees from outside, so a crash loop and an
eleven-hour silent web app passed it. Graduates research 032 on the operator's word.
mesh/merge-gate pass: the change touches no module of the mesh's graph
mesh/repo-check pass: its merge-check.sh passed
A walk mesh-delivery never lets go waited for ever with nothing open; the
controller now says it itself (S16), and the delivery's own stalls are its
conditions too (D14, H2 through close).
A build registered before its turn is carried by the next send to its
machines, so a published delivery asks nothing until its walk starts; the
rows the build needed (appeared, adopted, held on no walk, held then go)
and Phase B's verbs-without-probe are now said.
mesh/merge-gate pass: the change touches no module of the mesh's graph
One owner for one commit's journey, so 'did my change go out' is answered by
one module instead of five records joined by hand; delivery groups make the
cross-repository order the mesh enforces instead of the person merging.
Turning the merge check on for every repository showed it asked the wrong questions: a
repository chose whether it was checked, the gate mapped a change onto modules its own
way, the shared-code rule rebuilt 103 modules for a root script, and nothing kept a
commit off the trunk from becoming a module's version. Records the operator's decisions,
narrows ADR 0237 decision 4, revises to-be 45 §9 and Phase 5 and to-be 30, closes issue
280's left-open, and gives this repository its own merge-check.sh.
An image is not byte-reproducible, so ADR 0236's no-move rule never held
for one; the rule now also reads a build's source. Progressive insight on
ADR 0236 says what the rule assumed and what stands.
The operator approved Phase 5. Decides what the design left open: the build seat runs the
merge check, the facts live in the artifact store, the merge gate composes the mesh as it is
and with the change and judges only what the change adds, a replay lives where its incident
is, and the controller's tests run a bus of their own at the mesh's release. A core issue now
resolves only with a replay or a stated reason, checked by cycle.py.
The release gate judges a module from outside (applied, no new condition,
tools answer); a container that restarts in a loop or a service that fails
its own work passes it. Investigate a health declaration in every manifest.
Planned bus upgrades and rotations after one leak raised healer-wanted,
because the exemption was keyed on two causes. Progressive insight: the
exemption is read from the verb table; the decisions stand.
The 103-module rebuild ADR 0236 put down to the build agent came from a file of the catalogue's
reference module, read as shared because its definition was not in the merge. Records the issue,
corrects ADR 0236's open question as a progressive insight (the tiering was right; the build agent
never widens a plan) and amends to-be 30's rule for what a merge changed.
The coordinator's review: at the switch to roll, every send would carry the old default's
backlog unjudged. Measured: 88 of the 103 rebuilt modules were byte-identical, and no
build waited on any machine.
Phase 4 of to-be 45, started by the operator: the core's health as probes, a gate on
every plan's first machine, the rollback there once per build, the witness's contract
with the host, the bus as a step a person starts, and — with those in place — rolling
out as the default, keeping record where a module says why. 47 pushes by hand in one
day are what it ends.
Record the single-sample flaw found across the probes and watchdogs, and
the summaries that carried addresses and paths past the operator channel's
content rule; amend to-be 45 §4 with the two-look rule and the summary rule.
The operator decided the two consumers that took a failed write must retry and never lose an event: record how (a thrown write, a spool that takes the last delivery and replays, idempotent writes, a bound that borrows max-deliveries), and why the module counts its own deliveries.
Resolves ADR 0233's flagged item (the bus's streams copied as live files) as
a progressive insight pointing here. The bus module's own account is granted
the snapshot API and nothing else; restore builds a new store beside the live
one for a person to swap in. To-be 43 gains the bus's section and its restore
steps; design 25 and to-be 45 point to it.
A refresh answered with an empty body was read as JSON, so the media server module's handler threw after scanning; each finished download was offered five times and raised a false max-deliveries. Records the one rule for taking an event, and where the fixes are.
Nothing large is walked; the first night's size on the home server is measured and stated, the
previews are left out, the platform database is dumped, and the bus's live copy is flagged.
The operator's direction after issue 273: what data a module keeps, how precious it is and how it
is protected is said once, in the manifest, and backups, sticky bindings, retirement on unassign
and the self-check's watch are derived from it. Amends to-be 18, 32, 43 and 45.
The operator found four gaps that would bite on first use: a lost phone
locked the mesh's only person out, an operator message had no addressee,
asks could not name the specifics they need, and the operator was a
holder's flag. Each is now a rule with its check and its build phase.
Research 028 graduates: the mesh needs to ask as well as tell, over
channels that are seats rather than code inside one notifier, and an
action a person must approve cannot rest on a desk click an agent can
forge. TOTP verified by the controller is the proof; a key stays optional.
Amends to-be 45 section 5 as ADR 0227 left it to.
The operator widened the direction: asks need not be about permission, the work context picks
the channel, and desktop buttons should serve at the desk. Make the conversation the core model
and keep authorisation as a layer on top, with proofs a desk click alone cannot give.
The operator asked to approve and reject through Telegram, for a generic shape in which a
channel holds a seat, for input triggers, and for capabilities that decide which actions travel
where. Widen 028 rather than open a new effort, since its Q1, Q7 and Q8 own the question.
Issue 258's seat rule, written for the resolver, re-bound a machine's database consumers to the store
holding the seat elsewhere, where each was made an empty database, and nothing said so. Record the
incident, limit 258's rule to provisions that keep nothing, and decide that a binding to data is
kept and moved only by a pin.
Two modules existed only to say one thing each: networking required mesh-wireguard and nothing else,
and public-acme pointed the one proxy at Let's Encrypt. dhcpcd and cloudflare-dns are assigned
nowhere. Retire all four, keeping every machine's network and every certificate as they are.
Five passes are 25 s, shorter than a real hiccup; and a TypeScript provider's
remove can destroy, so retiring never calls it. Every TypeScript provider is
mark-only until it gains a retire that disables.
The operator's decision replaces ADR 0229's hourly release: the mesh no longer
finishes alone what its bound exists to question. To-be 45 and the module
protocol amended; 0229 and 0224 point to where their mechanism moved.
Phase 3 of to-be 45 was built: the details its section 7 left open are
decided in the record and the design says them as built — where budgets are
counted, the settle, what a healer may send, H4 on D6's consumer-behind, the
report verb's shape, the mesh-wide brake and S15.
Phase 2 of to-be 45 met questions its paragraphs do not answer: a strict
node-engine refuses an unknown key, a lease bucket can be raised again from
nothing, a command at a shell sends declarations too, and the bus's grant
is composed by the controller that needs it. The answers, the wire contract
and the withdrawal brake go into to-be 45 with issue 270's Phase 1
decisions; issue 272 records that the SDK's provider loop says nothing on
the bus.
The controller read 0113 as refusing to replace any value given to it, so
two leaked secrets only their own modules read could not be rotated without
a person handling a new value. Amends to-be 13 and points 0113 at it.
The operator-channel's holder said three hours of cleared conditions as new
the moment its consumer was made; recorded so the class has a home, with
its fix in mesh-catalog #88 and the to-be 45 §5 amendment still owed.
Building Phase 1 of to-be 45 found four of its rows depend on later phases
and S9 hears only the controller's own connection; the decisions the build
made that the design does not state are recorded for an amendment.
Phase 0 is being built in mesh-controller and mesh-host. Granting the
controller its own buckets showed that its grant never covered the work
queues' cancelled sets, so every cancel times out.
The operator approved the conclusion: the core must say when it is wrong
instead of waiting to be looked at. Record the nine rules with their
checks, and specify the parts and the six phases a build can take.
Issue 263: one global 20-character bound held keyless provisions to an
object store's key, was found only when a provider composed, and then
refused the provider's whole machine. Take ADR 0049's option C, check
overflows before merge, and leave an overflowing consumer out of its
provider's grants instead of refusing the provider.
The operator's mandate: races, ignored commands and silence keep recurring.
Classifies 48 recent issues by class, proposes nine checkable principles and
a phased roadmap, for review before anything graduates.
The same empty schemas recurred for push, plan, assign, pin and settings;
for push the dropped machine became a push of the whole mesh. Record the
evidence, the cause in the console's address form, and the open fixes.
The identity provider's admin lost the mesh's password again when its database
moved, and 31,000 silent failures followed. Record the recurrence, the rule
that makes a failing provider visible in status, and the module's self-repair.
The build of both parts is in review; the designs now say how they work and
are checked, and ADR 0223 records that the managers' own DNS mechanisms could
not write the mesh's file.
musl asks every listed nameserver at once and takes the first reply, so ADR
0196's public fallback answered NXDOMAIN for mesh names in every Alpine build
on the home server. Decide two mesh resolvers and no public line now; record
resolv.conf moving to the uplink's holder and /etc/hosts with /etc/hostname
moving to one hostname seat as the next steps. Amend to-be 08 and 26.
Issue 190's remaining steps: the runtime's module states the registry trust through
${seat:mesh-artifact-store:reach}, the private network stops writing it, generated resources
meet the collision check, and the rollout is an order rather than one push. ADR 0082 and 0102
get notes saying where their mechanism now lives.
A named push's cascade sent every machine a build held back by `record` or by
a plan waiting on its first machine, so a change meant to be walked through
the mesh one machine at a time reached all of them at once (issue 259).
Records the decision, narrows ADR 0083's flush with a dated pointer, amends
to-be 30, and locates issue 259 in the controller.
The rule that keeps resolv.conf the mesh's was checked by nothing, and a
retired seat and an unused module still read as live options. Amends to-be
26 and 08 to match.
The operator asked for tools to see, cancel, clear, stop, pause, continue,
restart and replay builds; none existed. Queue actions are the controller's,
process actions the build seat's per machine, and every action that drops
work leaves a failed outcome so no plan waits on it. To-be 18 amended.
ADR 0218: grants before code, one machine first, a newer merge takes over an
older plan (to-be 30 amended). Issues 250 (a merge announced twice), 251 (the
record's checkout owned by another account), 252 (a merge's changed modules
read wrong), 253 (the collector would delete every kept archive; to-be 18
amended, ADR 0189 corrected as a progressive insight), 254 (plans run over
each other); 249 located.
248, located: a consumer made with the server's default replays the whole
stream, and the controller's held every new merge and build behind a week of
old ones. 249, open: a module's new state reaches its bundle before the
grants that let it use it.
The nightly collector is back on the store (mesh-catalog#57). It was out for
a day because while-stopped named the module-local id and the host refuses a
declaration naming a container it does not have, whole; the namespacing was
fixed the same night and the composition was read before anything was sent
this time — plan's distribution.collect shows while-stopped as
distribution.store. novox applied it with no refusal and the registry was
untouched.
The store is 40G at 14:37, the filesystem 56% full; the collector first fires
at 03:30. Deleting a manifest frees no bytes until then, so that is the
number to read against.
244: mesh-controller.plan and .node publish an empty schema and then refuse
with 'needs node', including when node is passed — the console drops what
the schema does not declare. A tool that cannot be called is worse than one
that is absent, because it is listed. Worked around through the login shell,
which is the path the console exists to replace.
Graduates research 029 and amends design 36 (section 8): skills, subagents,
commands, hooks and output styles in one nox-mesh plugin; servers, settings
and instructions in the managed files; mesh, node and home scopes.
The operator scoped backups to mistakes, not disasters (issue 242). Issue 238: fix the failed logins
at their source instead of exempting the operator's address.
241: the SDK harness read a failed read as "no consumer" and withdrew all seven databases on the
control node; withdrawal destroyed data in seven providers. Fixed in mesh-sdk 0.1.10, mesh-catalog#44
and mesh-host#22; the recovery and what it lost are recorded. 242: nothing backs anything up.
Six of seven checks confirmed and autoMode from managed settings very likely;
the agent cannot undo its own managed settings, which a design must allow for.
The operator's direction: mesh, node and home. A plugin carries the first
two; the home scope places items the module owns path by path; instructions
follow the same layering.
Skills, subagents and hooks have no machine-wide place but a plugin, and
hand-copied homes have drifted and gone stale on every machine. Records the
evidence, what the vendor allows, and the options a decision must settle.
Rules for what the agent may do were set by hand per machine, invisible to
the mesh, and a session cannot loosen its own permissions; design 36 now
takes them as a module setting under the mesh's own keys.
239: two repositories defined photos; a rebuild to the catalogue's main replaced the app with a stub
and nothing refused it. 240: a dry-run build was recorded and its definition reached the machine.
238: the forge refused three ssh logins as the operator's account in 31 seconds; nothing in the mesh's
ssh configuration names the forge, and no jail ignores the mesh's own public addresses.
What runs since 2026-10-04 and what its first live use showed: refused
requests as timeouts, a late machine reading the whole set, the partial
secrets guard; the agent module and the licence manager on every machine.
225: a grant secret is composed with the account that reads it — the node's
account for a bundle, the declared secrets-owner for a container. Zero EACCES
since 12:30:10 where there had been 4330, both users created, mongodb logging
Authentication succeeded for each. The harness also stops calling a permanent
refusal a race, which is the half that cost three hours.
226: normalising moved to the records, where the provenance is known, and a
reference the sweep will not address is skipped rather than ending the sweep.
The first build after the roll-out collected 200 and said 1126 remain — the
backlog falls with every build instead of standing at 1681 for ever.
227: the three photo modules publish the endpoint they declare, and a
catalogue-wide test makes it a rule: a container publishes only a port its
module declares, or the mesh has nothing to assign and the number escapes.
232 came out from under 225: photos asked for a database its user does not
live in, invisible while no user existed at all.
Found fixing 225: with the secrets readable the provisioner created both
users at once, and photos still could not connect because it asks for admin
while its user lives in its own database. The password was never wrong — the
grant secret and the consumer's environment hash identically.
One fault wore the other's clothes: while no user existed anywhere, the
error was a complete account of 225. Worth keeping as a habit — fix the
first and look again.
Eleven node seats; a display as a provision with the machine's reach; other
modules' lines through the tool's own drop-in directory or ADR 0204's slots,
now also for xinitrc and xresources; the display server's module writes the
session's start.
A service needs node-service-manager held on its node, a package
node-package-manager, a container node-container-runtime: derived from the
resources, never stated; refused at assign, reported at composition until the
three holders are on every node. Glossary: depends on a seat; nothing claims a
package.
The operator's flow: clients publish what their credentials file holds, the
manager takes in a licence it does not own and rotates it from then on. The
token itself cannot be published (design 32 §10, ADR 0201), so a node reports
fingerprints and identity as state and hands the grant over only when the
manager asks; adopting is refreshing, newest login first; bindings with a
generation replace the rotated/switched events. Designs 36 and 39 and to-be 40
amended; a pointer note on ADR 0183.
ADR 0173 §2: a module is what it declares, and there are no kinds of module. The two records called
a resource under a home and a module placing one home-scoped; the wording is corrected in place,
marked and dated, the decisions unchanged. Design 36 and to-be 40 now say the module declares
/etc/claude-code and ~/.claude as directories, so the ownership check sees both, and declares no file
under either (mesh-catalog #244).
The dated note on ADR 0183 now rests on ADR 0193 and 0198 rather than on a bundle having no way to
call: the manager starts every exchange by the operator's direction, through mesh/ask. To-be 40's
WP4 no longer waits on a record — ADR 0198 is it — and the live proofs count the console's five
tools (ADR 0195).
Design 38's WP1-WP4b ran: the node's tool runtime is live on all four machines as the operator
account, tools are bundles given only their declared words, and a bundle has no bus credential.
So the wait on design 38 WP3 is over, the agent module calls nothing and the manager starts every
exchange (key, hand-over, waiting login, reconcile), and the manager's daemon now waits on WP4c's
record instead. Accounts are stated on all four, sudo -n works for each, the agent is installed on
all four; the plan's WP0 shrinks and WP2 gets a configuration-only live proof before any licence.
Designs 36 and 39 say what is built; this says in which order and what proves each step, in the
shape to-be 38 gave the operator's machine. Seven packages: the operator states the facts (accounts,
roles, licences); the console provides its endpoint; the licence manager and the agent module are
built and unit-tested in parallel; the manager goes live on the control node; the agent on one
workstation, with the switch and the predecessor's files removed as the proof of the whole; then the
rest of the nodes and the retirement of the two catalogue modules built on the old placement. The
live proofs wait for to-be 38's WP3, because both modules' tools run in the node's tool runtime
(ADR 0175) and a per-module tool container would rebuild what that record retires.
Evidence from both workstations and all four machines, read-only, and the
questions each must answer: seats and gating for the display stack, who starts
the session with which environment, contributions beyond shells, sway as a
sibling session; the container runtime with docker-compose on workstations
only, software outside the official repositories, secrets in the account's
environment, and three security findings.
ADR 0203: the account's environment is one module's (seat node-environment);
every module contributes variables and PATH entries, rendered by the
controller as a POSIX file and as environment.d.
ADR 0204: shell code is contributed to the login shell in named slots, and
login-shell becomes the mesh's node-login-shell.
ADR 0205: software the distribution does not package ships as a pinned
archive of the module.
Issue 225: undeclaring a user stops a node applying; the shell is never
given back or checked.
To-be 41 carries the work packages; to-be 38 WP5 points to it.
The operator's proposal: one module, holding a mesh seat node-environment, is
the only writer of the account's environment. It renders contributed variables
and PATH entries as a POSIX file shells source and as environment.d for the
graphical session. The shell's contribution shrinks to shell code; the shell
render-then-service-manager-again positions become options weighed against it.
Opened after the zsh module's rollout (to-be 38 WP5) was stopped: every machine
carries the same predecessor-written startup file, the module's block would
duplicate it and drop lines, nothing installs the prompt, and execute never
reads .zshrc. Weighs how modules contribute environment and shell code, where
the operator's own lines go, ordering, and what a contribution is addressed to.
The derived-value record is renumbered a second time: the key-value-buckets
record took 0201 while this waited to merge, as the bundles record took 0188
before it. Both times free when chosen, taken by the time it landed. cycle.py
caught it; three repositories cite this record, so the number matters.
225 — a provisioner has not been able to read its grant secrets since 01:30,
when a module's own code left its container and the files stayed root's. Four
thousand refusals, each worded as patience, and two consumers unserved. Not
from ADR 0202 or 0189, which landed hours later; dates in the report.
226 — the store's sweep stops at the first reference recorded with an address
and collects nothing. A guard that cannot tell 'I will not ask about this'
from 'it would not answer' stops the wrong amount of work.
227 — the photo app's admin client asks for the port the proxy holds. A module
pinned months behind carries everything its branch gained, the first time
anything makes it move.
ADR 0201 gains the boundary found reading it back: a consumer keeping several
holders of a deriving provider is refused, because the two ends have no way
to agree. ADR 0189 gains two consequences — the sweep is bounded because it
runs inside a build, and an apply arriving mid-window reopens it.
That last one is issue 224, recorded rather than fixed: the host's rule for a
stopped container is to replace it, and while-stopped is the first thing that
makes a stopped container intentional. Both candidate fixes are decisions with
their own cost. Nothing is worse than it was; the store has never collected.
The bundles refactor took 0188 on main while this waited in a pull request,
and the mesh's own code cites that one, so this record moves. Only the number
moved; the decision is the one taken on 2026-10-02, and the record says so.
Issue 202 re-checked against the refactored main: the fault stands, and the
test that surfaced it now fails one step earlier on issue 203's new credential
guard. Proven again past both — mint the credential, compose twice, and all
eight of dnsmasq's resources appear only with the setting set. ADR 0164 is
noted as the decision that answers half of it, and is not built.
Issue 108: the artifact store has never collected anything. Fifty-three
repositories on the machine that serves everything else, and the only outcome
of leaving it is a full disk reported as somebody else's failure.
The mesh decides what may go — from its own build records, so it never names
a digest it did not put there — and the store reclaims the bytes in a nightly
window with its server held still. Deletion on the one door takes nothing a
push did not already have.
Designs 18 and 20 amended; issue 108 resolved.
Also issue 202, found running the controller's suite: a module whose required
setting nobody set is left out of the machine in silence, and dnsmasq became
that module this morning.
Issue 124: a value the mesh's own rule produced reached neither end as a
statement. The object store's provisioner derived each consumer's bucket in
its own code; all three consumers transcribed the rule into their own
definitions, one of them wrong, and each of the three also named the machine
it happens to run on.
A served value may now name the consumer the mesh is serving. Design 27
amended; issue 124 resolved.
Events miss a machine that joins after them and replay history where only the
latest matters. A module now declares state it owns and reads; the controller
creates the buckets, the runtime serves them on the bundle's channel. Designs 32
and 25 amended; grants measured against a running server.
The per-node resolvers 0194 retires held two kinds of names that are neither nodes nor routes: the
lab's scenario machines and an operator's own lines. A zone a module declares is forwarded by the
mesh's resolver to that module; /etc/hosts is held per node through node-hosts-file, the operator's
lines in its kept region. Research 023 parks seats that define what their holder owns.
ADR 0194 rejected sending every query to the mesh's resolver because a node with its tunnel down
would resolve nothing; a public resolver listed second answers exactly then. That drops the
systemd-resolved stub and the runtime's dns: containers copy the machine's resolvers. Narrows 0194;
amends connectivity §2.
Every resolution fault found on 2026-10-03 was a per-node copy disagreeing with the truth: a hosts
file read once, an operator's old line beside the mesh's, a node's resolver lent to a LAN. Every
tunnel already converges on one node. Retires node-dns-resolver for a mesh-scoped mesh-resolver;
nodes route only the mesh's suffix to it. Narrows 0121; amends connectivity §2 and the seats.
A progressive insight: the rule and its check were stated as a suffix test; the mesh composes both
names of a route and publishes its internal one. The decision is unchanged.
Publishing every routed public name at a private address turned ace's LAN-facing resolver into an
outage for non-members: a phone got the control-node's tunnel address for the mail server. Routes
have internal names since 0151 and the proxy certifies public names publicly, so nothing needs the
private answer. Narrows 0066 and 0151; amends connectivity §2 and §5.
Found preparing WP4, before the first module's bundle was loaded beside the runtime's own:
proven with a two-copies probe, located in mesh-tools (node-tools), fixed by mesh-tools #32.
Design 38 WP4 records it and the two things the package left to the module — escalation
through sudo, and the filter file read from the manifest's path rather than a container env.
The build role was a mesh seat with one holder and its worker a push consumer with one delivery in
flight, so thirty-five images rebuilt serially on one machine while three others idled. The bus was
drawn for the alternative — a seat's accept subjects on a queue group of holders (design 25) — and
0190 uses it as one pattern for every role: a node seat with accepts, every holder pulling one ask
when idle, the asker addressing the role. Building is the first use: node-build-agent, held by the
build-agent module on every machine with a container runtime. Notes in 0121 and 0162 where their
facts went stale.
The operator's direction, absent from every record until now: the SDK must not limit who writes a
module; tools and services may be written in any language; one module may ship several bundles
(tools, a seat's implementation, a daemon); skeleton first, a full implementation when the work
requires it. ADR 0175 had the runtime import a bundle, which only JavaScript can be.
0187 makes a tools bundle a process the node's runtime launches and speaks MCP over stdio to —
the vocabulary the runtime already speaks outward — so any language with an MCP library can write
one today and the mesh's SDK per language is thin; the transport stays in the runtime (0039's
refusal, kept). Importing a TypeScript bundle is the shortcut, not the contract. Notes in 0175,
0039 and 0150 say where their mechanism moved; design 38 records WP1 as built and adds WP1b (the
launcher and the skeleton SDKs); the glossary's bundle widens.
The predecessor's agent module was retired and its six files stayed on both workstations telling
every session to use tools that no longer exist. This is its successor's design, revised during
review on the operator's directions: the host is module-agnostic, the controller has no part, and a
real licence manager hands out the correct licence in every situation.
- ADR 0181 (reconstructed): the operator account is a node fact stated by the operator; the home is
derived unless stated; a resource may be placed under it owned by the account; a node with no
account refuses one. What the controller shipped on 2026-09-27 without a record.
- ADR 0182: inside a home the module owns the directory and the files it places, writes into the
tool's own files for its few keys, never declares a credential's content, and holds everything
else as found; a predecessor's leftovers are the operator's to remove once.
- ADR 0183: claude-licence-manager holds the mesh seat anthropic-licence-manager and owns the
Anthropic licences, grants (encrypted with a key the vault made for it), bindings per touchpoint,
usage and audit; one rotation source under a lease; a token travels module to module sealed to
each node's module key on request/reply, never as an event; the agent module alone writes what
the agent reads; the host delivers package and state and knows nothing else. A bounded exception
to ADR 0113; dated mechanism notes on ADR 0050 and 0113.
- To-be 36 (claude-code): the mesh's part of the agent's configuration lives in the agent's
machine-wide managed directory, owned whole by the module and written by its code; nothing under
the home but the credentials file of a subscription licence; the API-key licence through the
key-helper; the console as a node-scoped provision (to-be 34 amended); MCP servers as settings
with an mcp_configure tool; one agent directory per machine shared by every session.
- To-be 39 (claude-licence-manager): store, the two licence kinds, keeping a grant alive, the
hand-over, who gets which licence with the predecessor's fallbacks, adoption with the identity
guard, the seat's verbs.
Numbers taken across main and every open branch at the time of the merge; to-be 14, 29 and 34
carry dated notes; the glossary gains "operator account".
The runtime work of design 37 broken down the way design 28 broke down the
bus: what exists measured (the host needs no change — a process and an
archive are what the runtime and a bundle are; the runtime does the job for
one module and must do it for a list), the order the dependencies allow, and
eight packages each ending at a proof on the live mesh — the lab skipped by
the operator's decision, ADR 0149 cited. WP1 the runtime serves many modules;
WP2 the controller composes one per node (a node principal, bundle archives,
the runtime's process, the gate); WP3 the runtime is a module and the console
its serving mode; WP4 the packet filter moves first; WP5 the shell on a
server; WP6 the service manager on a workstation; WP7–8 named and not broken
down.
Main carried two records numbered 0175 and cycle.py refused it: the one that
landed last (the found front end is uninstalled) takes 0180, the next free
across main and the open changes, with a dated note; design 08's citation
follows it; the index is regenerated.
Every configurable thing on a node is a module, the home included, and a
module is whatever it declares (0173, extending 0040). A node varies a module
only through a setting rendered into the file or a kept region, never an edit
(0174, extending 0011; issue 168 first). One tool runtime per node serves every
module's tools on the host side, never in a container; the console is its
serving mode, renamed node-tools (0175, extending 0150; 0047/0150/0152 carry
dated notes). The login shell is a node seat held by one shell module with
`execute` as its contract (0176). A unit may be user-scoped and the service
manager is a node seat held by systemd (0177).
To-be 37 is handed off in-progress to mesh-host, mesh-controller, mesh-tools
and mesh-catalog, with the build in order: the account on every node, the
runtime, zsh, systemd, then the graphical stack. To-be 29 keeps ~/.ssh and
points at 37; 33 §6 and 34 are amended; the glossary gains node tools, bundle,
kept region, installed/holding, and retires flavor.
The predecessor is retired on every node and what it still owned on the two
workstations — some thirty modules of dotfiles, user units and /etc files — is
owned by nothing. To-be 29 covers one directory under the home; the operator
wants the whole machine, system folders and home alike, as modules: one
default configuration each, varied per node by settings or a kept region,
never an edit; roles the machine has once as node-scoped seats with tool
contracts; the graphical stack gated by a capability so the same catalogue
serves the servers.
Four documents: the intended behaviour in the mesh's words; the predecessor's
desktop measured (34 modules, one with 88 files, 4 flavors and ~90 theme
variables) against what the records already give and what is missing (the
account is empty on every node, no user-scoped units, settings leak, tools run
in a container per module per node); the direction the operator set for where
tools run — one executor per node, host-side, module-agnostic, the console
renamed and moved out of its container, superseding 0047/0150 for tools; and
the candidate seats of the environment with first verbs, the shell first.
Another session's 0169 landed first. The collision check from issue 155
covered issue folders only; it covers decision records now, and would have
refused this.
Designs 33 and 08 revised. The first node-scoped seat with verbs: rules,
reload, remove; the nftables module holds it from a runtime with NET_ADMIN,
the first container to declare a capability.
The bus was public only so a new machine could enrol before it had a
tunnel. The machine now makes its tunnel key first, the token is issued
for it and makes it a peer of the hub, and enrolment happens over the
tunnel.
Designs 08 and 05 revised; 141 resolved by ADR 0140 and 084 by ADR 0102 and
issue 128, both by reading; 143 and 144 decided, built on the matching
branches in mesh-host and mesh-controller, resolved when the home server's
record names the predecessor's chain.
Issue 191's route proxy needs to know who the mesh is to serve an
internal name correctly, and the first fix had it work that out alone.
The membership on the bus now carries it, from the same list the filter
uses. ADR 0138 gains an insight that the proxy is where internal reach
is kept; designs 08 and 25 say how.
The status warning names 49 users; 48 are modules that never speak on the
bus, and the one real fault, a declared broker secret filled with a
generated value, looked the same as the rest.
Rule 5 of ADR 0163 (a former target is removed) met issue 162 (an archive
has no removal) in the host's own archive, the first time a host carrying
former targets replaced itself; every machine applied nothing from then on.
192: no provision says where the console is, its port was never assigned, and nothing
owns a person's agent configuration since the predecessor left. 193: the query verb wraps
the caller's text in a read-only transaction the text can end, and its rows come back
keyed by BEGIN.
The record gets its built note; designs 05 and 09 the revisions; 086, 098,
099, 100 and 101 stay located because every machine is converged and the
record's live row — a take read on an adopted machine — has not been run;
090 is built in part, its network difference left for the take to say.
The controller merged §1, §2 and the composed ssh config on 2026-09-27 while hq still
called the design proposed. Record what is on main, what is held on a branch, and what
has no code — and that the account fact shipped without a decision record, which is the
next thing to write. Also drop the real account and node names the public repository must
not carry, and correct ADR numbers that drifted in a renumbering.
Proposed for the operator's review: settings declared with defaults and cost (0164),
container-runtime as a kernel capability (0165), node-container-runtime seat with the
host creating containers through its holder (0166), and the runtime's file written by
modules that are not its own (190).
mesh-controller PR 176 and mesh-catalog PR 198. Designs 27 and 18 carry the words: places,
accesses, ${access:<id>}, the default a definition still holds while the catalogue converts.
A setting overrides a key a contribution or served fact declares and adds none; a provider that must
tell its consumers an operator's value declares it as ${setting:…} (173). The mesh's own files for a
module are a placed directory, `place: "mesh"`, and forty-eight definitions name no host path for
them (174). Registration refuses a definition naming an installation, the day the list emptied
rather than a release later (0155, progressive insight; 134). Designs 27 and 18 carry the rules.
Design 33 in progress against ADR 0154 (the twelve verbs, the prerequisites built); design 35 for
the records module under ADR 0153, extending 0025; as-is 07 rewritten to a mesh that keeps no store;
as-is 12 and 13 updated; issue 006 built and waiting on its live check.
The work order's group-3 question answered: an ordinary module the mesh assigns to the machine a
person sits at, holding a minted credential, calling tools under a manifest grant (invokes), serving
MCP on loopback. Design 34; pointers in 33, 25 and 0095; module check designed into 12 (issue 148);
README stops claiming an indexing nothing provides (issue 006).
Reported by the operator: ssh by the bare name logs in, by the mesh name
is refused. The predecessor's generator writes the bare name only; the
mesh's ssh-client roster already matches both and is not yet shipped.
Found and fixed the afternoon ADR 0148 landed: mailu-admin lost its
database behind Mailu's own resolver. Two catalogue PRs; an insight on
0148 that a container's dns is a decision, not a preference.
Issue 110's cause was not the filter: the runtime had never been told,
and the resolver dropped a query arriving on a bridge. ADR 0148 step 3
landed once it did (109, 151 resolved). ADR 0151 composes a route's
internal name under the serving node and drops the suffixed alias
(139, 157 resolved). Design 08 amended; a fact in 0148 corrected.
Assigning postgres on ace claimed mesh-store and made novox's own store
assignment unresolvable. The resolver reads a manifest's claims as 'does
hold'; ADR 0110 says the assignment holds, by a deliberate act the controller
already has (seat …, HoldSeat) but resolution does not consult.
Hosts first, then the controller — a build and a push each, now that the
mesh delivers the host. The host refuses a lower sequence than it kept
and drains a batch by sequence rather than arrival; the controller
numbers each send under the node's hold, inside the signed bytes.
Measured: two pushes, sequence 2 in the kept declaration, counters in
the store agree, no machine reads as behind. That last one is the
subtlety: the mesh compares the digest of what it would send against
what it did, and a number changes the bytes, so the read-only comparison
composes with the last number sent rather than a fresh one.
nfs-share and smb-share share an intent, not a contract; one seat would be
a union with every field optional. What the exemplar exposes is that a
seat declared inside one module cannot be implemented by another without
depending on it.
All four machines run a host the mesh built, published and delivered, the
last delivery unattended: each stood aside once for a genuinely newer
version and the delivered launcher started it. A following push that
delivered nothing new was applied and reported by every machine and stood
nobody aside.
The crossover needs one restart of the unit per machine, once, because
the running launcher executes from its own inode. Measured timing: three
seconds on the machine, 17-20 as the operator sees it, the difference
being the control plane composing before it sends.
107 is unblocked: a declaration field is now a build and a push.
The operator's call: a node-scoped system seat family (node-nfs-share,
node-smb-share, …) defined by the control plane, one per protocol as
package registries are (ADR 0109), so several modules occupy it and a
machine may hold both.
ace exports the operator's media library over NFS and Samba with host
services no module declares: they close at converge, nothing owns their
configuration, and no module elsewhere can require the share. Proposes a
file-sharing module that accesses the paths, holds a node-scoped seat it
defines, and provides file-share to consumers.
Asked whether a newer host was delivered using the link-time stamp, which
every delivered host carries as 'development build' now that the version
comes from where the binary sits. Never matched, so it stood aside on
every push for ever; standing aside cancels the report, so the mesh never
heard from it. Read as healthy throughout.
The three-minute push wait made it invisible: a wait long enough to
absorb a whole apply is long enough to hide that the machine never
answered.
164 a credential that must be accepted is minted anyway
165 one accepted value must be accepted once per consumer
166 a requirement cannot be optional
167 code several modules share has no home
168 a setting reaches every file and every contribution
161 resolved and verified on a machine: the workstation runs a host the
mesh compiled, published, delivered and started, applying declarations
and reporting the version it was delivered as.
The system it was built for comes from the artifact — the one thing a
toolchain takes from a module, which 0142 already allowed because the
target is a property of the artifact. The version comes from where the
binary sits, which 0142 decided and nothing had implemented.
Two mistakes on the way, both caught by reading the output rather than
the line that claimed success. A second -ldflags does not merge with the
first: the binary gained its system and lost -s -w, 12.2MB against 8.5MB.
And the delivered binary was named after its package, so the first
delivery was correct, reported success and was invisible to the launcher.
A delivered host that cannot apply is a machine the mesh cannot repair,
because the declaration that would fix it is the one it cannot apply. The
launcher's fallback is what made that an inconvenience instead of an
expedition.
162 is new and not about the host: an archive has no removal, so a module
using one can never be unassigned, and the attempt takes the whole apply
with it — the machine applies nothing else either. It is how undoing the
first delivery froze the workstation.
The loop closed on the workstation: the version landed, the launcher was
replaced, the running host stood aside, and after one restart the launcher
started a binary the mesh had compiled, published and delivered.
The launcher goes as a file resource rather than inside the archive, and
that is the safety rather than a preference. A file is written atomically,
so the running launcher keeps the inode it started from; an archive writes
in place with truncate and would cut a script a shell is reading. The
manifest carries a second copy and a test refuses any drift from the one
in packaging.
Then it would have refused the first declaration it was asked to apply.
The Makefile links in two facts the mesh's toolchain does not, on purpose,
and one of them is the system the host was built for — read before
anything is applied, so the failure is safe and total. Nothing reports it:
the unit is active, the bus link is up, and the log says it is hearing
what the node should be.
Worse, the declaration that would fix it is the declaration it cannot
apply, so the mesh cannot repair such a machine. Restored by moving the
delivered versions aside and letting the launcher fall back, which is the
fallback working as designed.
0142 already settles the version — it comes from where the component sits,
not from its linker — and that is unimplemented. The system pin has no
answer, and the candidates are a decision rather than a fix: put it in the
path too, carry it in a file beside the binary, or stop pinning at link
time at all, which is 0005's to change.
ADR 0110 decided each assignment records where its requirements are answered
from; the control plane keeps only a per-machine pin (none recorded) and
resolves every requirement implicitly. Harmless with one provider; a second
one silently moves consumers' data.
Filed as a to-do. Nothing is broken by it: every machine here is amd64 and
reports so, and one architecture is enough for now.
When a machine joins, the mesh should collect what it reasonably can about
it and refresh that daily. It already asks what a machine can do; what it
is made of is the same question one level down.
More is already collected than it looks — eight capabilities, the links
that face outside, the host version, and on an adopted machine what it
holds, what is reachable and the firewall and tunnel it was found with.
The architecture and the kernel are in there too, and nothing reads
either: measured, all four machines report amd64 and linux, and node show
prints the capabilities beside them without printing them.
Missing: memory, disk, the processor beyond its architecture, the
distribution and its version, virtual or physical, cores, uptime. Several
are what somebody wants when deciding where a module goes, and the
placement code's own comment already imagines them.
Also missing: the refresh. A machine publishes after an apply, and the
five-minute reconcile publishes nothing, so the mesh's picture is as old
as the last push. Same mechanism 087 wanted.
This is the third thing in one day found to be collected and read nowhere,
after held resources and the host version. Whatever gets added should say
in the same breath which surface shows it, or it will be the fourth.
159 gains the note that the architecture is already reported, so matching
an artifact to a machine needs no new fact — only the comparison and a
compiler told what to target.
Asked whether the host is built for more than one architecture. It is
not, and the reason is worse than a missing feature.
A bundle in a compiled language must name a system, must name one of
alpine, android or arch, and is refused with a careful message if it gets
that wrong. The field is then read by nothing: it does not reach the
compiler, no machine is matched against it, and nothing chooses between
two artifacts by it. The compile runs with no target named and produces a
binary for whatever the build machine happens to be.
The host is x86-64 because the build machine is, not because the
declaration said so. Correct for this mesh by coincidence — four machines,
all x86-64 Arch.
A module declaring two systems would get two identical binaries, both
published and both pinned, and the one sent to the machine it was not
built for would fail at exec. android is the sharp end: not an x86-64
platform, and an artifact declared for it today would be an x86-64 binary
wearing the label.
A field that is checked and ignored is worse than one that does not
exist, because the check is what persuades you it works.
Also noted: the processor is a second dimension the manifest has no word
for, so even implementing the present field would not answer the question
that found this. And since the Go toolchain builds statically, one binary
would run on all three systems anyway — so the pin is a policy rather
than a necessity, which is a decision and not a fix.
Both of the things ADR 0141's insight named as remaining are built. A Go
toolchain based on a new mesh-tools-go module, so the compiler is named
and not pinned; and ${version} in any value of a resource that uses an
archive or a bundle.
Measured rather than asserted: the mesh built the host through its own
toolchain, published it to its own registry, and the bundle fetched back
out is a statically linked stripped binary that runs and says it is the
host.
The cost was larger again than 0141's note said. Three more things in the
path assumed one language or one shape — an entrypoint became a .ts file
whatever the language, the output directory was the compiler's to create,
and a bundle was refused if it named what it is built from — and a fourth
was in the base image, which is Alpine where the first Dockerfile ran
apt-get. That last one is issue 136 in an image, and the build refused
rather than a module failing later.
The version in a path is the digest, not the commit: two builds of one
commit are the same bytes, so a content-addressed version keeps the path
an unchanged build already had.
Still nothing delivers a version to a machine. The host module declares no
resources, so the bundle sits in the registry and no machine is asked to
take it. 0141 carries the insight and 142 the account.
It reported "N machines run an older host than another" by comparing
versions as strings. A host reports its version as a commit, and commits
have no order. On the live mesh it named the three machines running the
NEWER host as the ones behind — ced54d4 sorts above 04a27ca and that is
all it means.
The code even carried a caveat saying versions compare as strings and that
this "is enough for the timestamps and commits this mesh uses". That was
the error, written down and not noticed: enough for timestamps, meaningless
for commits, and the mesh reports commits.
It now reports the split and claims no ordering, which is more useful as
well as more honest — the reader sees who is on which side, and that is
what decides whether a field can be sent. Ordering is left with the host,
which would have to report something ordered for anybody to have it.
This is issue 145 arriving by my own door an hour after I closed it: a
report that confidently says the opposite of the truth is worse than one
that says less.
Measured live. All four machines run the identical host binary — same
digest, installed within eighteen seconds — and at first only one reported
a version, which read as a difference where there was none.
A machine publishes a report after an apply. The five-minute reconcile
publishes nothing, because it is the machine keeping itself as declared
rather than answering anything. So a current, idle machine never says, and
the mesh cannot tell that from a machine running something ancient.
Confirmed by pushing: not reported, then 04a27ca.
Enough for the purpose, not enough for the claim. For deciding whether a
new declaration field is safe it is sufficient — pushing is what the mesh
is about to do, and the answer arrives with the act. For knowing what the
mesh runs it is not, and "not reported" is worded as "nobody has asked
recently" for that reason.
Making a heartbeat carry it would close the gap and would change what a
heartbeat is — a bare word that the node is there, deliberately carrying
nothing else. Left alone rather than widened in passing.
145, partly resolved. The sentence that was true for eleven hours of a
mesh in which no module could reach another now says what it is not a
claim about: that is the mesh and the machines agreeing, and nothing here
dials a provision. It checks nothing and does not pretend to — ADR 0146
decides the check and is deliberately not built. What changed is that the
report no longer implies otherwise. Stays open for that reason.
Carried forward: 0146's check needs an internal name fetched over TLS with
the certificate verified, and until today no machine trusted the mesh's
authority. Three of four do now, so whoever builds it does not have to
solve that first.
107, diagnosed and deliberately not built. The premise is confirmed in the
host's own words — unknown fields are refused because "a field the host
does not know is a thing the control plane believes it asked for" — so the
fix is a flag day, not an addition. 087 now makes the cost measurable, and
the measurement is why it waits: one machine of four runs an older host,
it is parked, and nothing delivers a host at all (142). Shipping the field
means hand-placing binaries and unparking a machine, and one missed in
that sequence is unreachable, not degraded. The fault it prevents has
never been observed.
142 gains the note that it is 107's gate, and that it is what makes a
declaration field cost a rollout instead of an expedition.
A judgement about order, not a refusal, and cheap to overrule.
The machine has reported its host version since ADR 0141, whose own
comment says why it must: without it nothing can say a machine is behind.
The controller's copy of the report did not have the field, so it
unmarshalled into nothing and was thrown away on arrival. Two structs
describe one message and only the sending side had it.
node show names it per machine, "not reported" where the mesh has not been
told. status names every machine running an older host than another does,
and which is newest.
Disagreement rather than staleness, deliberately: nothing delivers a host
version yet, so the mesh holds no canonical current one and "behind" has
no fixed point. What it can say is that the oldest host in the mesh is
what the mesh may send.
Two refusals to guess: a machine that reported nothing is not called
behind, and versions compare as strings — right for the timestamps this
mesh uses, wrong for a scheme where 10 sorts before 9, said at the place
that would have to learn.
107 gains the note that this is what makes its new field safe to consider,
and that one machine of four is behind today, so it is not free yet.
Two of the four surfaces the report named already carried it — the host
has reported Held since ADR 0100, and node show reads the machine's own
list with an `as of` beside it. Recorded as checked rather than assumed.
Two did not. The apply line counted what it applied and said nothing
about the difference; status read the mesh's take-time listing, so a
module assigned after it showed nothing at all.
Both now say it, and the part that carries the weight: a hold suppresses
"all doing what they were told, all heard from, running what the mesh
would send them". That sentence was true for the whole outage, and acting
on it is what stopped the predecessor's proxy. Being adopted still does
not suppress it — a mode somebody chose is not a half-finished action.
Status does not call a hold a fault, deliberately. It is correct
behaviour, and a reader trained to see red for something the mesh did
right stops reading.
Extended after the first was proven. Every converged machine now holds
the anchor and verifies an internal name with a plain client; before,
the two unassigned ones answered 'unable to get local issuer
certificate' and held no entry for the mesh.
ace is excluded on purpose: it is adopted, so a module assigned there is
held rather than run, which is right and is not trust.
Both the resolution and 0147's insight said one machine of four, which
was true for about twenty minutes.
Registered ca-trust from the catalogue — it was merged and had never been
registered, which is why "assign it to one machine" had no module to name
— assigned it to the workstation, and verified.
Verified in the form ADR 0147 prescribes, against the authority's own API
so the handshake needs nothing else in the mesh to be right: 200, issuer
Mesh Internal CA, Verify return code 0. Four routed internal names verify
too, and `git ls-remote https://…` works, which is the consequence the
report named.
Removal exercised for the first time. Unassign and push removes the
anchor, empties the trust store of the mesh's authority, and returns the
plain client to the original error; assigning again restores it. That is
the half 0147 claimed and nothing had shown.
One thing it found that is not in the module: removal works only because
the host removes the service before the script. Stopping the unit is what
deletes the certificate and refreshes the bundles, and it needs the script
to still exist. The symmetry rests on an ordering nothing states.
0147's "written, and not yet run" now carries a progressive insight
saying it has run, where, and that it ran on one machine of four.
Verified on the machine: umami has zero restarts, applies all 26
migrations, reports the store up to date, and umami.novox.be answers 200
where it had answered 502 since 2026-09-25.
It was the same fault as 135, filed three days earlier and diagnosed
without either record noticing the other. 135's container IS umami — it
is named in 135's own evidence table, holding novox.internal:10.42.0.1
after the overlay range moved. The dial appeared to succeed and the first
real query timed out because the name pointed at an address that no
longer existed; the store, the path and the credential were all fine.
118's own reasoning is kept as a warning, because it is careful and
wrong: it argued path-MTU and conntrack, and it named the move that would
have found it — compare umami against a container created that week —
and did not make it. A stale name presents as a network fault, which is
why ADR 0148 stops copying names into containers rather than detecting
when the copies go bad.
Also: 118's located-in named mesh-catalog modules/umami, which was never
at fault; it names mesh-host's comparison now. And 135 gains the pointer
to 118, which is the back-reference I have now missed three times.
The certificate is genuine, from Mesh Internal CA, and nothing on the
workstation trusts it — verbatim the error the report gives. The public
name on the same proxy verifies cleanly, which puts the fault exactly
where the report puts it.
What is in the way is not an assignment. `ca-trust` is merged in the
catalogue and has never been registered with the mesh — 39 of 76
manifests are — so there is no module to assign. It dry-runs clean and
needs no artifact built.
Two findings from reproducing it, both their own issues:
157 — every routed name is published with an `.internal` alias that
nothing serves. The hosts file says keycloak.novox.be.internal; the proxy
serves keycloak.novox.internal and refuses the other by name. The first
three names I tried came from the hosts file and failed with a TLS alert
rather than a verification error, which pointed at a regression that had
not happened.
158 — the proxy re-logs all 52 routes every two seconds, 31 times a
minute. The one line that explained 157 sat between two of them.
Also recorded, because it was nearly filed as a defect and is not one:
step-ca publishes roots as /roots.pem, which is PEM, so ca-trust's fetch
and its refuse-a-non-certificate guard are both right. Its other endpoint
/roots returns JSON that contains the text the guard greps for, so the
guard is sound only because of which path is published.
Three records were left describing a mechanism a new record had moved,
and a reader arrives at them by following a citation: 0066 still said a
routed name is written into every container after 0148 replaced that with
resolution; 0016 still read as though the lab were the test bed after
0149; and issues 109 and 135 said nothing about 0148 ending the copying
that 135's own fix made comparable. Each was a citation leading to the
wrong answer in a record that was not wrong about anything it decided.
This is the second time in one session. The playbook rule I added last
round did not stop it, so the convention is now written where the record
conventions live, with the shape to use and three worked examples — and
with the honest note that it is NOT machine-checked and cannot be from
`extends:` alone: 102 records extend another, 87 have no back-reference,
and that is correct, because extending usually means building on a
context. Making it mechanical means a record declaring the relationship
in frontmatter, which is a schema change and is not mine to decide.
Also: designs 18 and 20 claimed `updated:` dates from before I edited
them, and 117's `fixed-by` gained the commit beside the record.
**0114 accepted.** The separation it draws — a consumer's resource and a
credential that reaches it are different things — is a data-loss rule,
and a record naming one should not sit unresolved while the code that
could hit it is written. Not built, and accepting it schedules nothing;
accepted-and-not-built is where 0141 and 0142 already are.
**0068 superseded by 0149, the live mesh is the test bed.** Never built,
and contradicted by practice that was written down nowhere but a handoff
note. The faults that cost the most are faults of a mesh that already
exists — bound consumers, containers made against an older roster, an
adopted machine — and a bed is by construction a mesh that does not.
Issue 156 settles it: that change was verified honestly, on the only path
where it cannot fail. The lab is not retired; raising a mesh from bare is
now its whole job.
**117 answered by 0150.** A module's own code runs as supervised
processes under the module's one account. 0047's argument was for a
runtime per module, not for a container — a unit satisfies it and runs as
an account. The invariant is the account, not the process count: its
worry was a second identity to scope and seal, and processes sharing one
account create none. Designs 18 and 20 now cite it and 0047 carries a
dated pointer, which closes the third disagreement — that neither design
knew the record existed.
Answers issue 151. Copying the roster into each container made the roster
part of each container's identity, so one name moving replaced every
container in the mesh — and it never stopped the staleness it was for,
since a copy taken at creation is stale the moment the roster moves (109,
135).
A container resolves through its machine's resolver instead, and nothing
is copied. Staleness stops being possible rather than detected, and a
name's blast radius becomes nothing.
Scoping each container to the names it binds was the close call and is
rejected: it contradicts anything-calls-anything, and leaves the roster
in the digest so the churn returns for a widely-bound name.
Gated on issue 110 — a container on the runtime's default network has no
DNS at all today. Removing the copy first reintroduces 109 and 135
silently on a live mesh. 151 stays open until the code lands; design 08's
file-not-resolver passage is narrowed to the machine's own roster.
Three issues named the branch that fixed them, and a branch is deleted
when it merges — so every `fixed-by:` was a pointer that resolved to
nothing by the time anyone followed it. They name commits and pull
requests now, and playbook 03 says to.
Two records were missing the thing a reader arrives for. 146 did not say
that one of its fixes crash-looped the control plane on a running mesh,
which is the whole reason the delivery subject carries the stream and the
raise path was the only one exercised. 151 did not say that 152 removed
the false reasons its roster moved, or that it stays open for the real
ones.
ADR 0080 enumerates what cycle.py enforces and named four things; it
enforces five. A progressive insight names the fifth — the decision
stands, the list had gone stale. The checks README and playbook 03 gained
the same rule, and 155 points at all three.
Measured on the broker's own config after the fixed controller composed
it: every node is now allowed _DELIVER.<node>.> as well as the bare
subject. That grant was missing when this was diagnosed, which is why
re-making the consumers would have silenced every machine.
Also records the five consumers it kept and reported, including the build
machine's worker, which is bound the same way and was not anticipated.
146's change is correct on a mesh being raised and fatal on one that is
running: the server will not move a push consumer's delivery subject
while a subscriber is bound, and a node is bound to its declaration
consumer the whole time it is up. Reproduced before fixing; an earlier
version of the test unsubscribed first and passed against the broken
code.
The consumer that works is kept and the assertion says so. Not re-made:
the nodes are granted the bare subject only, so re-making would have
silenced every machine.
Two machines filing issues in the same hour both read main correctly and
both took "the next free number". main lags every open pull request —
seven that evening — so they collided twice. The second collision
reached main with records, cycle and index all reporting success.
cycle.py now refuses a tree where two issue folders share a leading
number, and names both. Proven by adding a duplicate and watching it
fail. The colliding records become 153 and 154, renumbered in the branch
that lands last, because renumbering a branch whose author is still
pushing only moves the race.
The check catches a collision; it does not prevent one. Taking a number
still means reading the open pull requests as well as main — issue 155
says so.
It ran 22:46-23:11, five full replacements, and stopped when a pass
happened to read the store during a window it was up. The record said
nothing about it stops; that was wrong. Exiting by luck is the finding,
not a mitigation.
The three gatherers pass over a node whose set does not compose, and
raise anything else. 151 stays open: this removes the false reasons a
roster changes, not the fact that a real change still replaces every
container.
ace numbered its two records 147 and 148, which this repository already
uses; they become 150 and 151, as 127 became 149.
The new record is why the control node could not stop applying: the
roster alternates between two values because routeNamesInTheMesh
swallows a per-node plan failure, and the roster is part of every
container's identity. The loop closes through the control plane's own
store, which each pass replaces.
ADR 0138's reach is internal, public or both. ace's LAN devices (an IoT
switch on mosquitto, unifi access points, plex clients) are none of them;
the only reach that admits them is public, which states the wrong thing and
opens the port to the internet on any machine with a public address.
ADR 0112 decided that an assignment places a directory and says where an
access's data is; the control plane implements neither. ace's media stack
(a 40 TB ZFS library, plex state on a second disk) cannot be expressed
without writing ace's paths into manifests.
Both found migrating the first module on ace (searxng), adopted with route-adapter:
147 — assign is not a no-op for a routed module: its route reaches the predecessor's
proxy while its container is held, and the name serves a dead backend until take.
148 — every push that moved the mesh's names replaced every container on novox, twice,
the control plane's own store included.
It asked a question rather than reporting a defect, and the question was taken
five days later — the mesh's own components are binaries on the machine, and
third-party software stays a container because an image is the right way to
carry somebody else's build. The delivery of them is issue 142.
A number identifies a record, and two were given 127 on 2026-09-27. The one
three documents and three source files cite by number keeps it; the other
becomes 149, says so in its own heading, and its two inbound references are
repointed.
It is also resolved: an empty declaration is sent carrying owns_nothing rather
than skipped, and the host refuses an empty body that does not carry it, so
emptiness cannot be read as a truncated declaration.
0037 (where a module lives), 0113 (the vault makes every secret) and 0115 (one
assignment of a module per node) described arrangements the mesh has — a
catalogue of other people's software plus a table filled by module add, a vault
answering a secret requirement six modules make, and a rule the assignment
table's primary key already enforces. Each is accepted against what was built,
and says so in its own words.
0037's other half is not built: a manifest outside this catalogue has no check,
which is issue 148.
0068 (the lab takes requests) and 0114 (a shared credential rotates over two
credentials) stay proposed. Neither is built, and both are decisions rather
than records of something that happened.
The record catches README claiming this repository is indexed into a knowledge
base. Nothing indexed it, and since the cut-over there is nothing to index it
into — the surface that answered is on the transport the mesh removed (issue
147). Noted where the record is, so the next reader does not go looking for a
search that cannot exist.
088, 089, 120, 128 and 130 each name a commit that is on main and cites them —
the forge's address following a moved port, a route naming its endpoint, a
provisioner asking the backend what is there, the hosts file written into a
marked block, and undeclaring giving a unit back the state it was found in.
Each says it was closed by reading commits rather than by a run, so nobody
reads a green that was never measured.
129 stays located on purpose: ca-trust is merged and no machine holds it, so
the symptom it opened on is still true everywhere.
Written 2026-09-28, answered the same week by mesh-controller bdf965d and
c68d3a7, and never closed — so the mesh's own account said reach was declared
nowhere while three readers were reading it: the filter, the proxy's names, and
each authority's host policy. The resolution names them and what checks each.
What is left is retiring the older per-port keys the block replaces, which is
not a gap in what reach can say.
Diagnosed from the configuration: the tool server is HAL's brain, a local
process on the workstation with the predecessor's broker URL in the assistant's
own config. No manifest, no assignment, no seat, no account. Nothing regressed
— the mesh removed a transport this program still dials, and the program was
never part of the mesh. The mesh has a tool model and nothing publishes an
operator-facing surface onto it.
Every tool call fails with 'AMQP not connected', on every node including the
local one, because the tool surface still opens an AMQP connection and that
transport was deleted at the cut-over. The mesh reports healthy throughout —
what broke is the thing standing outside asking it questions, so nothing the
mesh checks is about it.
Not about enrolment. A push consumer delivers onto an ordinary subject and
everything subscribed to it gets a copy; the controller's two consumers were
both named after it, so both were given the same subject and the one process
acted on every message twice. Enrolment is where it drew blood because a second
enrolment mints a second credential.
Two more faults behind the three already fixed. The account a token is the
password of was never recorded, and the comment above the issuing code said
it was; issuing now records it. Placing the composed list at genesis is the
other half — the control plane says what it composed and whoever raises the
machine writes it beside the bus, because no declaration can reach a machine
that has not enrolled.
With that a first node enrols. It is then enrolled twice from one attempt,
each minting a credential, and it keeps the answer to the first while the mesh
keeps the second. The trail for that one stops at a duplicate that survived
message-id deduplication.
Raising a first node hits them in order: the bundle's bus image named for a
registry that is gone; the certificate made by openssl in an image that has
none; enrolment dialling TLS at a bus that speaks first, then refusing its own
token for the empty bus. Each was right until the bus changed and nothing has
raised a foundation since.
The fourth is not a patch: the installer carries the bus's first user list and
the controller composes the rest through the bus being a module, and at genesis
there is no module — so the first node cannot be let onto the bus it just
raised.
Found trying to run ADR 0147's bed. The older bundle raises the previous
broker and a control plane that refuses to start without MESH_BUS_NATS; the
newer one stops a step earlier, asking for a certificate from openssl in an
image that has none. No bed can run while this holds, so 0147's check section
now says what actually stands behind it — the rendering, not a machine.
Issue 129: every internal HTTPS name fails verification on every machine,
because nothing has ever written the mesh's root into a trust store. The
report proposed the controller inject it the way the private network writes
the registry's trust; this record rejects that — reachability and trust are
not the same fact, and where anchors live is the host's difference, not the
controller's. A module requiring internal-acme-ca does the whole of it, and
being unassigned undoes it.
0145's core stands — a module checks what the mesh claims, from where the callers
are — and everything it said about what to dial was wrong.
A raw port is not how anything in this mesh is reached. A real caller resolves a
name, the proxy answers, the proxy reaches the service; dialling a port tests the
last hop of a four-hop path and skips the three where most connectivity lives.
So each hosting form gets its own endpoint, route and name:
connect-docker.<node>.internal for a container, connect-process for the mesh's own
code in a unit it writes, connect-unit for a unit a package ships — and the same
under each public domain. Every machine checks every machine, by name, over TLS,
verifying the certificate against the authority that should have issued it. The
names are the instrument: a failure reads as connect-docker.g14.internal did not
answer.
No name is written anywhere: the machines come from the roster fact, the labels are
the module's, and a machine that joins is checked by the others on the next push.
Five things this needs that the mesh does not have, named rather than assumed —
a module every machine has (ScopeNode is exclusivity, not obligation), a container
running a bundle, a unit a package ships, the public domain in the roster fact, and
issue 129, until which every internal name will correctly fail verification.
The mesh asserts three things are callable and has never checked any of them. A
module on every machine serves an endpoint of its own and dials every other
machine's, from the position the callers are in — not the host and not the control
plane, both of which reach those addresses by paths no ordinary caller uses and
would have passed throughout the outage.
Its probe is its own endpoint, declared reachable over the private network, so it
is admitted by exactly the rule that governs every internally-exposed service. The
tempting target is a service every machine has, and those are the ones never closed
— ssh above all — which would have passed for all eleven hours.
Resolution and connection reported separately, because they have different owners.
One failure is not a fault, and the count travels with the result. It reports and
repairs nothing.
Built on a scheduled container, a rendered roster fact and a bundle: no new
vocabulary. Adopted on its own merits, which is what 0143 was not.
Everything should be able to call what runs on the same machine, another machine's
service exposed to the private network, and another machine's service exposed
publicly. Three cases; the filter had two.
The first was expressed as the machines' own addresses on the private network. A
caller on the machine carries such an address; a caller in one of its containers
carries a bridge address and matched nothing — measured, same destination and same
machine: src 10.10.0.1 against src 172.17.0.8. The second case worked by accident,
because the tunnel rewrites a caller's address to the sending machine's. Two of
three working is why it read as correct.
0143 answered the wrong question. It proposed verifying each grant from the
consumer's own network position and went to length about which position, because
whether a caller sat in a container changed the answer — and that difference was the
bug. Observing a configuration error is not its remedy. Superseded, and nothing
replaces it; whether the mesh should check a grant is still open in issue 145 and
must stand on its own.
And a module is not a container: 61 of 72 happen to use one, 11 do not, and a rule
reasoning about containers describes most of the mesh rather than the mesh.
A grant is four facts and a credential — the provision, the machine, the port, and
who the consumer is when it connects — and it is the whole mechanism by which
anything in the mesh reaches anything else. The mesh asserted it and never found
out whether it was true. Issue 145 is what that cost: eleven hours of 'every module
current with its source' while a module could not reach its database.
The consumer verifies it, from its own network position. Not the control plane,
which reaches the address by a path no consumer uses. Not the machine, whose own
packets carried a source address the filter admitted while every container's was
refused — so a host-side check would have passed throughout the outage it exists to
catch. That is inference from the rule that was loaded, stated as such.
A connection and nothing more; speaking each provision's protocol would be a second
implementation of every provision. One failure is not news, only consecutive ones,
and the count is reported rather than the last attempt. A consumer that is not
running reads unchecked, which is a different sentence from broken.
What it costs to be wrong is the constraint on all of it: a check that calls a
working provision broken trains a reader to ignore the report.
Converging the control node closed every path by which a module reached another by
the machine's own name, and it ran for eleven hours while the mesh answered 'all
doing what they were told, all heard from, running what the mesh would send them,
and every module current with its source'.
6,154 database failures in one module's log, beginning at the minute of the flip.
The service accepted TCP and never answered HTTP. Every check the mesh makes
passed, because every check the mesh makes is about the relationship between the
mesh and a machine — applied, current, containers running. None asks whether a
module can reach what it requires, though the mesh composes every grant and so
knows exactly who requires what.
The filter fault is fixed. The eleven hours are the measurement, not the bug.
Also records on 144 that 'more closed than the mesh believes' was harmless only
for what is reached from outside, and an outage for what is reached from within.
The record says internal and public each mean something to the filter. For an
endpoint the proxy serves, the second half is wrong, and ADR 0045 said so first:
a public service is exposed through the proxy, listening from the mesh, not by
opening its own port.
Found by trying to express one real module, not by review — routed name public
because browsers post to it, machine port private because it serves a dashboard in
cleartext. Under one value for both, saying public would have reopened a port an
operator had just closed. Measured the same evening: the routed name answered from
the internet over TLS while the port was refused from the same place.
Corrects a fact. One statement per endpoint with three things derived from it
stands; the filter column applies to an unrouted endpoint.
The first account said the declaration carries no resource that would disable the
found firewall and that nothing implemented the sentence the flip prints. Wrong:
the mechanism is a step in the host's own apply, retireFirewall, and it is careful
— it reads back that the mesh's table is loaded before retiring anything, records
the forward policies first, and verifies ufw reports inactive afterwards.
What is established: ufw was active and enabled two minutes after a flip that
reported the node converged with nothing failed; and the machine's record now
reads disabled_by_mesh: true, written by a reconcile fifty minutes later that
found ufw already inactive because an operator had disabled it by hand.
So the step did not take effect and the record says it did. The candidates are
named rather than chosen — the step is skipped silently when the apply has any
failure, the mesh's table is loaded by a service in the same apply so ordering is
open, and the host's detail lines do not reach the journal, which is why this has
candidates instead of a cause.
Both found by converging the control node — the first machine with a firewall to
flip, since the two before it had none.
143: the preview and the flip both say the found firewall is disabled. The node
reported converged, 372 resources applied, nothing failed, and ufw is still
enabled and active. A converged node's declaration carries no resource that would
disable it; the sentence is printed by the command and nothing implements it.
144: ufw was never what filtered the traffic that mattered there. Fifty forwarded
openings converged through it had matched zero packets, while a chain the
predecessor installed in the container runtime's pre-accept hook did the work —
in memory only, recreated by nothing. The mesh's filter now covers that path, so
the machine no longer depends on it, but the chain remains and is the only thing
refusing the bus and the registry, which the design requires reachable from
anywhere so a machine can enrol before it has a private address.
Measured: the host is a binary somebody copied to four machines, owned by no
package and built by nothing, while the controller, catalogue, builder and vault
are container images publishing no ports at all. Same language, same project,
same kind of work, delivered two ways — and the difference is not a judgement
about either, it is that images are the only delivery that works.
What it costs: genesis must raise a container runtime before the control plane
can exist; updating the control plane goes through a registry the control plane
runs; a host change cannot be rolled out at all; and compiling the language the
mesh is written in is not a capability of the builder, so the controller is built
from a hand-written Dockerfile — the incantation the bundle toolchain exists to
abolish.
Third-party software stays a container: the store, the registry, the broker are
somebody else's build. The container runtime stays on the machine for modules.
What changes is that the control plane no longer needs it to exist.
The receiving half is already built and tested (ADR 0141). Staged: compile Go, an
artifact names its target, deliver a binary, the host first, then the rest, genesis
last.
The record claimed a version reaches a machine as an ordinary archive with
nothing new needed. Two things it needs do not exist: no toolchain can compile
the host (the list is typescript and python, and the control plane, also Go, is
built as an image from a Dockerfile instead), and nothing interpolates a built
version into a resource path, so nothing can ask for .../versions/<version>/.
The decision, the options weighed and every consequence stand — the host half is
merged and tested. What was understated was the cost, so it is corrected in place
and dated rather than superseded.
The supervision was already right — a clean exit means the host stood aside and
the launcher runs what is on disk, failures are counted, and a rollback happens
at the limit. Two things made it dead code: nothing told the running host a
successor was waiting, and the rollback resolved its known-good version through
pacman, which no machine here uses and which two of three operating systems do
not have.
Keeping a version rather than a path was the clue. Versions live side by side in
directories named for them; the newest runs; the running one stands aside between
reconciles; a reconcile that completes records itself and retires what is older
than its predecessor; rollback starts that predecessor. No new resource kind and
nothing new on the bus — an archive already fetches by digest, and the path
written is never the path executing.
Answers issue 142.
A host change merged yesterday reached no machine without a person copying a
file. The host is not a build target, no declaration delivers it, and the half
that recovers from a bad host — noticing the executable changed, a known-good
record, a launcher that rolls back — is written, tested and called by nothing.
All four machines run a byte-identical hand-copied binary that no package owns
and no record names, so nothing can say a machine is behind.
Found because ADR 0140 needs the machine to report a new fact, and merging that
could not roll it out.
Reading a converged machine's rendered rules showed the cause: the chain blocks
everything passing through and then allows the machine's own containers back by
listing their address ranges. 0137 made that list typeable and 0139 tried to
generate it; both refined a list that should not exist, because the mesh has no
position on a container reaching outward. Constrain what arrives from outside,
allow what did not, and let the machine report which links face outside — one
fact instead of a list. Ports keep following the modules unchanged.
The records check now allows one record to supersede several, and stops
requiring a withdrawn record's own citations to be live.
Both follow from the same rule the mesh is built on — a node's configuration is
composed from the modules assigned to it. Reach was settled separately by the
filter, the proxy's names and the certificate authority, so "this must not be
public" could not be written; it becomes one value on the assignment that all
three read. And the forward chain consulted two constants plus a typed list
although modules already declare their networks; it now forwards what they
declared, with the host rendering the addresses it allocated.
Found preparing the control-node's convergence. Reach is settled independently by
the filter, the proxy's names and the certificate authority, so "this must not be
public" cannot be written and a public certificate is obtained regardless. And the
forward chain allows two hardcoded ranges plus a typed list, though the mesh
already knows which networks exist because its own modules declared them — a range
wide enough to keep four of them would have forwarded two predecessor leftovers too.
One container restarted 2286 times over five days while the mesh reported the machine as doing what it
was told. Its overlay address was five days out of date: the host compares a container by a digest of
its spec, and the mesh's names were not in it, so a container whose image and files never changed was
left alone holding a name that no longer resolved. Forty-eight others were current only because
something else had recreated them.
The same fault as issue 045, in the field that was left out. Resolved by putting the names in the
digest.
0120 was already accepted; what failed the check was that it rests on 0112, still marked proposed —
and so do four designs. The decision stands: a definition names no node, no mesh and no host path, and
everything a module needs is a requirement the mesh resolves.
Accepting it makes the gap visible rather than hiding it, so issue 134 states it. 0112 says how it is
checked — 'a catalogue test finds no domain name in any definition value' — and there is no such test.
Asked by hand: seven modules name this installation in a value the mesh acts on, and eight mention a
public name in prose nothing reads. The two are not the same fault and the fixes differ, which is why
the issue separates them rather than counting to fifteen.
Two of its statements are built — a version preparing its state, and the mesh saying what it applied —
so the document is in-progress rather than proposed, and names the code that owns them. Resting on a
live record rather than a superseded one: 0127 was replaced by 0131.
What this exposes is pre-existing: it also rests on ADR 0112, which is still proposed, and a document
that is not itself proposed may not. ADR 0120 has rested on it the same way for a while. Accepting or
superseding 0112 is a decision, not a cleanup, so it stays visible in the check rather than papered
over.
ADR 0135 made a step something the mesh derives for any module that prepares its state, which turned
ADR 0052's reach into a fault: a module whose database is briefly unreachable would stop every module
declared after it on that machine — the fault issue 011 already removed for every other shape, and
the reason the catalogue migrates itself at start rather than in a step.
A step now stops the rest of its own module and nothing else; an action still gates the machine,
because genesis is a row of them and they belong to no module. What was not attempted is reported as
skipped rather than left to be inferred from silence.
Two faults in 0133, both caught on review. It put the declaration on a container — one resource kind
the host applies — so every author would restate the machine's arrangement and a module's own
lifecycle would be tied to how its artifact happens to run. A module declares entrypoints for its
tools and its provisioner; preparing its state is the same vocabulary and nothing about a runtime.
And it derived the scope from the machine, which the facts already answer: a consumer is a module on
a machine (issue 022, migration 0015), so what the mesh provisions is per consumer. A module on three
machines has three databases, there is no shared state to race over, and the lock obligation 0133
invented was for a situation the mesh does not produce. The level question HAL answered with stages
dissolves — the scope of preparation is the scope of the state, and the mesh knows it.
0133 keeps its reasoning and gains a pointer; design 32 and issue 133 name the live record.
0133 — a module owns its migrations and the mesh owns when they run. A container declares what must
run before it; the mesh derives the gated step from the resource it precedes, so the image, the
environment and the credentials come from the one place they are described. The module owns the SQL,
the dialect and the lock; the mesh owns the moment and refuses to start a version whose step failed.
Per node, with no level: a step that ran once somewhere leaves every other machine ungated, and
'once, mesh-wide' is what holding a seat already means.
0134 — the pipeline is observable from a merge to an artifact and goes dark at the machine. What a
node now runs, and what it refused, become facts under the control plane's own seat, emitted when
what a machine runs changes rather than on every convergence pass.
Design 32's lifecycle carries both; issue 133 points at them as what ends the matter it opened.
The mesh replaced its own control plane with a build carrying a migration, applied none of it, and
then recorded no build for three quarters of an hour while saying everything was fine. ADR 0052
already prescribes the shape — a run-once step that gates the server — and the control plane was the
one module that did not use it.
A role's tools belong to the role, not to whichever module holds it today: the seat declares them
with their schemas, serving them is a condition of occupying the seat, and what the mesh can do
becomes a read of its own records rather than a question nothing answers. A module keeps its own
tools — the same module may run without the seat, and then only its own name is true.
Design 33 follows: the three families, addressing a node-scoped seat, discovery, and what serves
this to an agent.
Nine modules could not be rebuilt: their record named the repository and no directory, so every
build looked for a manifest at a repository root that has never had one. Resolved by mesh-controller
— `module add` takes the directory and the forge, and the rule is checked rather than described.
The AMQP transport is gone from the control plane and the hosts (mesh-controller #112,
mesh-host #39). On the way: no build had ever recorded its bases, so every bases-first order
walked an empty graph; the builder now reports what it was handed and the graph is read from
builds (mesh-controller #113/#114). Issue 131 is resolved by the forge module's merge event,
the control plane following it, and those edges.
Tasks 4.3, 5.2 and 5.4 are done as of 2026-09-28 02:25: every machine reports on
the new bus, the seat is held by the module that provides it, the old broker is
unassigned and forgotten, and every credential was minted afresh at the end.
5.2 records what it took, in the order it was found and each fixed on the trunk
before the next step, and how the bootstrap loop was broken once, by hand.
Until now the holder was derived — assigned and claiming — and a second eligible
assignment was refused, so a seat could not pass from one holder to the next
without a moment where nobody held it. The controller finds its own bus through
one of these seats, and that moment took the control plane down on 2026-09-27.
The holder is now a row the controller keeps, written by `seat <name> --to
<node>/<module>` in the same write that removes the previous one. No row means the
old rule, so nothing changes for a mesh that never hands a seat over; with a row,
another eligible assignment is silent rather than refused, which is what lets the
next holder run beside the current one until the switch. A holding is the
assignment's and goes when it does. Each rule names the test that checks it.
Under ADR 0131; design 28 task 5.3 is the work.
Taken during the outage of 2026-09-27, when the protocol leaked into the seat's
contract: to hold mesh-broker a module had to provide amqp, so the module that
will carry the bus could not hold the seat that names the bus, while the module
being retired could. Supersedes 0127. Modules depend on the seat and reach the
bus through the sdk; no manifest provides or requires amqp; the old broker's
module and the two modules that required it leave the catalogue; the AMQP
transport is deleted once every node reports on the new bus.
Design 28 step 5 rewritten under it: the seat handover becomes its own task and
is built first, because the seat the control plane dereferences cannot be empty
in between — that emptiness was the outage. The cost note now carries what was
measured rather than what was assumed.
0128 and 0130 extended 0127; each now rests on 0131 with a dated note and
changes nothing it decided. Every other citation of 0127 names its replacement.
records.py still fails on 0120/0112, which predates this branch.
Derived while merging and visible from no single repository, so it belongs
written down rather than re-derived later: the client library before the
catalogue, because it is where the subject is derived and a converted module
against the old one publishes the local name itself; the catalogue before the
controller, because the controller refuses an old-style name outright and would
make every unconverted module unregisterable.
Two tested properties are what make it safe and not merely ordered. An old-style
name passes through the derivation untouched, so an unconverted module keeps
working at every step. And a converted name derives to exactly the key the old bus
published, so nothing moves on the wire until 5.2 sets the variable.
The failure avoided is 127's own, which is why this is worth a table: a publisher
and a subscriber disagreeing about a subject log nothing anywhere.
Correcting design 26 to match what the merged code does, and fixing issue 112's
status, which used a word the vocabulary does not have.
A claim on a seat the manifest does not itself declare is refused at registration
rather than by the parser. A module may hold a seat another module declared —
that is why ADR 0126 has a caller name the seat and not its provider — so whether
the name exists is a fact about the whole catalogue.
And a declared seat may promise nothing. That is a marker seat, and most
node-scoped seats are markers: which module is this machine's packet filter. ADR
0126's "a declared seat carries a protocol" says what a holder must satisfy, not
that every seat offers something.
`records.py` still fails on ADR 0120 resting on a proposed ADR 0112, which is not
this branch's and not mine to decide.
Both lines of work numbered from the same point, so four decision records and one design
document existed twice with different content. The trunk keeps its numbers and this branch
yields — the only rule that scales, because the trunk's are already cited by what merged
before them.
0117 the bus is the only broker -> 0125
0118 a module declares its own seats -> 0126
0119 amqp is a provision, not the bus -> 0127
0120 the mesh bus is required -> 0128
0123 a seat carries its role's protocol -> 0129
0124 the predecessor is ending -> 0130
design 29, what a module declares -> design 32
Applied to the code repositories too, because a stale reference is worse when numbers
collide than when they dangle: the reader lands on a real record that decided something
else.
Two reconciliations the merge forced, both real:
**0110 was marked wholly superseded and was not.** Its successor says in as many words that
everything 0110 decided about what a seat *is* stands untouched — and two records that
landed on the trunk rest on exactly that part. So it is accepted again, extended rather than
replaced, with a note saying which of its claims moved and where.
**A seat's protocol becomes columns, not fields.** The trunk moved the seat set out of
compiled code into a table the controller owns. This branch had added what a role accepts,
emits and serves to the Go slice. The decision is unaffected and the mechanism is better for
it: giving a role a protocol is now a write rather than a rebuild, which is the trunk's own
argument applied to what this branch added.
One check still fails and it fails on main too: a record resting on ADR 0112 while that is
still 'proposed'. Left alone — it is not this merge's to answer.
ADR 0119 rejected giving the old broker a retirement condition and said why: "its
clients are not only the predecessor's, so the retirement condition describes a day
that will not come". The operator has said that day is coming — the predecessor is
deprecated, some of it still running, none of it being migrated, left to stop rather
than moved.
Recorded because three documents reason from the premise it overturns. Design 25 §5's
"no day anything is waiting for", §9's "the predecessor's clients never notice", and
design 28's closing note that the predecessor's world does not need to move.
**And it needs no new machinery, which is 0119 being paid off rather than revised.**
Because that record made the broker an ordinary provider rather than a compatibility
module, ending it is unassigning a provider whose provision nothing requires — something
the module system has done since it existed. So step 5.3 finishes instead of trailing
off, and the transitional double announcement of a build outcome has a date.
The consequence worth planning around: the predecessor's own mesh talks over that
broker, so shutting it down ends the tooling that reaches this installation's machines
from a workstation. The rollout is driven from the node, or driven before the broker
stops. That is a sequencing constraint on 5.2, not an afterthought.
What survives is `amqp` as a provision: a module that genuinely needs an AMQP broker can
still be given one. What retires is this broker's role as the predecessor's.
`rollout check` answers from records whether this mesh could move its bus, and names the
next step for each thing missing. The move itself waits on that check having been run
against a real mesh — writing the irreversible half before its question has ever been
asked of something real breaks the plan's own rule about beds by another route.
And the cost of being wrong is written down rather than assumed: the old broker stays for
its other clients, nothing in a served request's path goes over the mesh's own bus, and
what a failed move costs is the mesh's ability to change things rather than the services
its modules serve.
Three ways to do the replay were weighed and the right answer was that the bus being
moved to already does it. A queue on the old bus receives only what is published after
it is bound, so everything built before the catalogue existed was announced to nobody. A
stream is a log and a consumer is a position in it: a consumer created later starts at
the beginning and the builds are simply there. Checked against a running server, because
the decision rested on it.
So "who replays" has no answer because nothing replays. The mechanism was never about
builds — it was about a queue that could not remember, and carrying it across would have
carried a workaround for a limitation that no longer exists, with nothing looking wrong.
Retiring it belongs to step 5, with the rest of what only the old bus needs.
The account existed as a permission model and as nothing a person could be given; there
is a record and three commands now. The client is two surfaces over one thing, a command
line and an MCP server, both using the client a module's runtime uses — so what a person
may do is answered by the same permission list that answers it for a module.
Design 25 §7 says nothing of this is built before its bed passes, and this was built
before. Noted in the task rather than quietly ignored.
A node's intrusion filter should be composed from its assigned modules, like
its firewall (the Filtering mechanism): a service module (postgres, mssql,
mailu) declares its jail in its manifest (filter + stanza, no node/path per ADR
0112), and the mesh writes the jails of a node's modules into the fail2ban
holder's jail.d. The base (sshd, recidive, ignoreip=mesh-range) stays the
fail2ban module's. Records the model after novox's HAL per-module jails were
lost as dangling symlinks; the ignoreip is now safe on disk, the service jails
need this to be restored.
A foundation template that stands the server up, writes its settings and the mesh's
first user list beside them, and starts a controller on the new bus. The first user list
is the installer's because at genesis there is no mesh to compose one — a bootstrap
credential, rotated like the store's.
The carried list is checked against what the controller derives, since a mesh cannot be
raised twice to discover they disagreed. That check immediately found the composer
granting a role's whole event branch as well as the one event it follows.
What is left of 4.3 is running it, which is 4.1's bed.
Both sides behind a seam, one implementation per bus, and the outcome is the role's
own event so one publish reaches the asker, the controller and the catalogue. Checked
against a running server, including the part the decision rests on: a third party
hears the same outcome the asker does.
Reviews 0110/0121 after a session where renaming seats cost three freezes, a
builder deadlock, and hand-resolved manifests. The seat rules were right; the
set being a compiled Go slice referenced by name-string everywhere was the
mistake. Seats become a table keyed by a stable id; claims/held/production code
reference the id; a rename is one UPDATE, no rebuild, no re-registration, no
freeze. The build machine reads the set from the mesh instead of embedding it,
removing the controller/builder seat coupling. Closed set and scope naming
unchanged; only storage and reference change. Outstanding renames (registry
seats, private-network scope) wait for this — as data each is a write.
The mesh's own seats said who does a job and nothing about what may be said to
them or by them, and that gap showed up three times in one day looking like three
different problems: a build machine with three audiences for one outcome and no way
to derive a grant for any of them; an event genuinely about a role with nowhere to
live but the namespace of whichever module holds that role today; and a catalogue
catching up on builds, where every option needed a grant the design refuses.
One cause — the mesh has roles it cannot describe. So the `mesh-*` seats take the
same three fields a module's seat has, and the machinery that already derives
authority, queues and consumers from a declared seat does it for these too.
Builds become work submitted to a role, and `mesh.build.request`,
`mesh.control.built` and the BUILDS stream retire. A work queue shared by several
build machines is exactly what a seat's `accepts` is, so a second mechanism for it
was two places a permission could be wrong. The outcome is the seat's own event,
which means one publish still reaches whoever asked, the controller that records it
and the catalogue that places it — the fan-out a shared exchange gave for free,
written as a subject the mesh derived rather than a topology somebody configured.
That also avoids the grant that ruled out the alternatives: no holder needs
permission to publish into an asker's inbox.
The blocking gap is now named rather than incidental: the shared library has no way
for a module to publish on a seat. The build machine is Go and reaches the bus
directly, so it is unaffected; the artifact-store event waits.
Records the manual update process (module moved -> build -> reconcile; and the
breaking-change freeze/re-register recovery), and the two things that make
self-update more than a webhook: the build-on-push trigger is currently HAL's
(hal-gitea-tools on :9877), a retirement gap the mesh must replace with its own
forge-webhook trigger wired to every repo including mesh-controller; and the
builder validates manifests too, so a breaking change couples controller +
builder + manifests + hosts, and renaming the builder's own seat deadlocks its
rebuild. Names the transition discipline (accept old+new for one release) that
self-update needs so a push does not auto-freeze.
Every module named its events the way the old bus spelled a routing key, so on the
new bus every cross-module subscription pointed at a namespace nobody publishes to.
Nothing failed — the services started and none of them reacted. Converted, and the
rule now has checks at both scales: at registration for one manifest, and as a test
across the whole catalogue where a consumed event's emitter is present.
It was larger than the report said, in two directions nobody had looked. Forty-three
files of module code pass the event name at runtime, so the code mattered as much as
the manifests. And both clients had to learn the mapping — without that, converting
the modules would have broken the mesh that is actually running, which is the
opposite of what fixing this was for.
Design 29 gained three things it did not say: what a wildcard is (`*` for one name,
`**` for the rest, spelled the mesh's way and derived to each bus's own), that an
event about a role belongs on the seat and why that is not yet possible, and how the
rule is checked — because "a subscription that matches nothing is silence" is exactly
why nobody noticed thirty-seven manifests being wrong the same way.
4.2 and 4.3 are unblocked. The catch-up half of 4.5 is not: it is a decision, and it
narrowed rather than closed. It cannot be a reply to a module's inbox, because that
needs the blanket grant design 25 §4 refuses.
Records the reversal: distribution stays as the mesh's OCI registry (it serves
every artifact-store:// image); only verdaccio, a redundant second npm registry,
is removed. The 'consolidate onto gitea / retire distribution' direction was
dropped. Also records that the node-* rename was executed as one controlled
migration with a brief compose freeze, and why the delivering registry seats
are deferred rather than folded in.
The control plane's seats grew a second naming style (the-*) beside mesh-*,
and the closed set was the only place any seat could be defined. This settles
both: system seats are mesh-* (one, mesh-wide) or node-* (one per node), named
for scope; a module may define its own seat outside the closed set. Folds in
the seat review: mesh-build-machine (scope fix), mesh-private-network (one
server + client modules, dropping per-node VPN choice), showcase becomes the
first module-defined seat, node-uplink, and the node-* renames — plus the
registry consolidation onto gitea, which reshapes the registry seats and gates
retiring distribution/verdaccio. Records why the renames are a coordinated
migration and why distribution cannot be removed until gitea serves images.
Minting on both halves, the file delivered per push, and the bus's objects
asserted on every start — verified against a real server that asserting twice
changes nothing, that a machine joining an already-raised bus is accepted, that
each node's consumer is bound to its own declaration subject, and that CONTROL
does not dead-letter before the controller gives up.
Which bus the mesh is on is one fact, and being told about both is refused at
start rather than warned about: a mesh half on each is one where a declaration
goes out on one bus and the report comes back on the other while every component
logs success — ADR 0074's failure arriving through configuration rather than code.
So 4.1 no longer waits on code. Both links speak NATS, the composition happens,
and every claim behind them has a unit test or a check against a running server.
What none of those can stand in for is a mesh raising itself, which is what the bed
is — this is where the code stops and the lab starts.
Two corrections of fact, both found by building the module's image and connecting
to it as a host would.
The first composed configuration said `verify: true`, which makes the server
demand a *client* certificate — and nothing in the mesh presents one. A host pins
this server's exact certificate and authenticates with the password the mesh
minted, and so does a module's runtime. Every connection in the mesh would have
died at the TLS handshake before any password was looked at, with an error that
reads as a fault in the client. TLS is still required; verify only decides whether
client certificates are checked. Mutual TLS is a later question and would need
machinery the mesh does not have — a certificate per module per node.
And §4 read as though the controller wrote the whole file. It writes the user list
and nothing else: ports, TLS paths and a store directory belong to the container
the module raises. The two files share one directory of necessity, because an
absolute include path is resolved relative to the including file's own directory.
The decision stands in both cases — accounts are composed, not called for, and
passwords are minted and sealed. What changed is what the file says and who writes
which half.
What is in: a bus user's hash is recorded and its plaintext returned once, and
the user list is derived from the machines, what each runs, every manifest and
which machines hold a live token. Permissions stay derived rather than stored,
because a stored copy could disagree with the records it came from while both
looked internally consistent.
What is out, with what each needs, so the next person does not rediscover it:
delivery, which has one open question about what a module declares in order to
receive the file — design 29's ground, not this document's; minting, which is
transport-coupled because an enrolment reply carries one password and a node on
the old bus must not be handed a credential for the new one; and calling the
assertions from a start path.
A roster fact may be shared — written into a marked region of the machine's
file (into: block, hq 128) rather than as the whole file. The template
renders the content; shared decides how the host lays it down. Composes with
hq 128: the region mechanism is the host's, the format is the module's.
Correcting a tick and a claim I made one commit ago. 4.1 does not wait on an
enrolment user per live token; it waits on the whole composition, of which that
user is one input.
Tasks 1.3 and 1.4 are honest about what they built — the composer, the
derivation, the permission model, the stream and consumer definitions, the
asserter, all pure and held by unit tests and a golden composition. Nobody wrote
the caller. Measured: outside the package that defines them there is not one use
of the composer, the permission derivation, the stream set, the stream asserter or
the principal type. Step 1's "done when" claims every account and permission
composed from the manifests, and a mesh raised today would stand up a server with
no user list at all.
It also needs state the mesh does not keep. Design 25 §4 says the file holds
bcrypt hashes and that passwords are minted and sealed exactly as today — but
today the mesh mints one, hands it to the broker through a management call, seals
the plaintext to the holder and keeps nothing. With no management call the hash
has to survive every later recomposition, because the first thing a new module or
a person's access change touches is a file that must still hold every other
user's password. No bcrypt hash is stored anywhere in the controller.
Named as its own task rather than folded into 1.3, so the gap between "the parts
of step 1 exist" and "the mesh does any of it" is visible.
The facts mechanism formatted the roster in Go in the control plane — one
formatter per fact, in the consumer's own configuration language. ADR 0120
makes a fact a path and a template: the mesh owns the data, the module owns
the format, and the control plane holds no format at all.
to-be 29 (operator accounts + what lives under a home) is rewritten to ride
it: the ssh files become roster templates, the whole ~/.ssh is owned with a
found/owned boundary that cannot lock the operator out, keys are mesh-owned
through an SSH CA (existing keys adopted not regenerated, the operator's
personal key signed not minted), and the ssh-agent is a user-scoped service.
All three halves of the host's link are through seams, and the reply address in
the payload is now proved from both ends rather than one — the test asserts the
transport's own field held the consumer's ack address by the time the request
arrived, so a server that stopped claiming it fails a test instead of letting the
reason become folklore.
Two things had to be built for the host to hear anything at all: a node's
declaration consumer, which only the controller may create, and the enrolment
user's inbox, which design 25 §6 names and the composer granted none of. Both
were silent gaps — a node with either missing looks correct and hears nothing.
What remains is a single piece: something that composes an enrolment user per
live token. On the old bus that account is made imperatively through the broker's
management API; here there is no management API, so issuing a token has to
recompose the server's configuration. It is the only thing between the two links
and a mesh raised on NATS from nothing, so 4.1 now says so.
The controller's inbound is through a seam with both transports behind it, and
the store window is now the server's rather than the controller's memory. Seven
claims about that were asked of a running server rather than reasoned.
Wiring the controller's own subscription is what found issue 127: every event
name in the catalogue is still written the way a routing key is, so design 29's
derivation turns a consumer's declaration into a subject no emitter publishes.
Thirty-seven manifests, one that cannot be composed at all. It fails on the first
mesh raised on the new bus and not before, which is why nothing had caught it —
the conformance fixtures pin one emitter against one subject, and both halves of
that pair are correct.
The node-facing flows are unaffected: those subjects are the mesh's own and
derive from nothing a module declares.
It waited on a retirement condition 0119 abolished when it made the
deprecated broker an ordinary provider. A step waiting for a condition
nobody set would sit open forever.
A claim the whole enrolment handshake rests on, now measured against a
running server rather than reasoned from documentation — and held by a test
so it cannot become folklore if a server version changes.
The guarantee is the same and the mechanism is simpler — a nak with a
delay, no parked list, nothing lost when the controller restarts. It costs
one thing: a naked message comes back whatever happened meanwhile, so an
older report is redelivered after a newer was applied. A report already
carries the digest of the declaration it answers, so supersession becomes a
check rather than memory — ordering settled by what a message says, not by
when it arrived.
Two things. A paragraph from the superseded 0117 survived beside the 0119
correction that reversed it, so §5 said both that the amqp interface
retires and that it does not. The stale one is gone.
And the framing. §1 opened with "the bus carries five kinds of traffic
today, and this design keeps the five", with a column mapping each to the
queue it used to be — which describes the mesh's nervous system as a port
of something that did a fraction of this. It now says what the bus is: a
role addressable without knowing its holder, the mesh's own state, work
that queues until somebody can do it, and permissions derived from what a
module declared. Conditions, observation and a person's client land there
too as they are built.
Glossary gains `bus` and `the deprecated broker`, with a note on why not to
say "compatibility broker" or name it after a protocol — the second invites
exactly the backwards framing this commit removes.
The mesh models machines but not the people on them — a node record
holds no username, and no module places anything under a home. So who
you are on each node (jochens/ace/jochen) is unknown to the mesh, and
nothing owns ~/.ssh, dotfiles or ~/.config. HAL knew it; the nox mesh
dropped it. Proposes the account as a node fact and a home-scoped
resource class (the ~/ mirror of ADR 0112's /var/lib placement), with
the login key staying the operator's (ADR 0051). Not urgent — HAL's
generators still run — load-bearing at node-by-node retirement. Found
generating ~/.ssh/config from HAL's registry, which nox has no
equivalent for.
Task 3.2. ADR 0074's model is untouched — floor plus capabilities, partial
implementations legitimate, identity from the credential, dedup on
x-event-id, conformance as executable fixtures. The transport beneath it is
rewritten: exchanges and queues become subjects and streams.
Statements marked *verified* were checked against a running server while
the runtime's client was written, not reasoned from documentation. Three
of them are things the specification would otherwise have got wrong:
- the payload is the body alone, with metadata in NATS headers; an
implementation that nested the whole envelope would agree with nobody
- a durable name may not contain a dot, while the ack subject joins two
names with one — conflating them looks right in a permission list and is
refused as a consumer name
- a certificate must carry a name the bus is dialled by, because the NATS
client has no hook to replace hostname verification the way pinning did
on AMQP
And one limitation lifts: a module may now call another's tool. Issue 049
recorded that a scoped account could not declare the reply queue a caller
needs, and ADR 0095 routed every ask through the control plane because of
it. Per-account inbox prefixes plus allow_responses replace that. ADR 0095
is not reversed — the control plane is still how a person asks — but
module-to-module calling stops being a question about capability and
becomes one about policy, which `uses` already answers.
The NATS client has no checkServerIdentity hook, so pinning no longer makes
the name check redundant — the bus's certificate must carry a SAN matching
the address nodes dial.
A holding is derived at resolution from manifests, never stored, so there
are no recorded old names to rewrite. The work is an edit plus a kept rename
table — kept because a module lives in its own repository and may be
registered long after the catalogue stopped using an old name.
The manifest already uses serves for a provision's facts, so a module's
tools take their own key. Declaring them is itself new — until now a
module's tools existed only in a runtime environment variable.
Storage is not a property of a subject — a stream is a separate object that
covers one — so the question is always how many streams, not which topics
are durable.
Three facts decide it, two of them verified rather than assumed: NATS
refuses overlapping streams instead of merging them, so a shared stream
plus a per-module one is not available at all; a filter cannot express an
exception; and a stream per module turns one cross-module consumer into one
per module. So one stream, with per-subject caps for the fairness that
matters. Per-module age is genuinely unavailable, and a module that needs
it declares a seat.
2.3 was already true and is now proved — the seat refusal is generic, and
three tests pin what matters: a second bus is refused by name, a different
bus implementation is refused too (which is what makes the bus replaceable),
and the AMQP broker no longer contends so both run on one mesh.
2.1/2.2 turned out not to be a no-op. The host keeps a container only when
its spec matches exactly; genesis raises the upstream image and the module
declares the mesh-built one carrying the entrypoint, so assigning it
recreates the container. That is ADR 0067's pivot and it is safe only
because the bus carries nothing yet — which is why step 2 comes before
anything speaks NATS. After it, never again: the config is a directory
mount, so accounts change without touching the container's spec.
Design 29 said no module requires the bus. The catalogue disagrees: 49 of
72 modules take a broker credential and 23 do not, so an ambient connection
mints an account for a third of the catalogue that never speaks — and the
49 each hand-write the path it lands at, which is provisioning done badly
by hand.
The bootstrap argument that made it ambient was narrower than it looked.
"A provisioner needs an account before it can run" is true of a provisioner
process and says nothing about a provision the controller answers, and the
controller is not waiting on a bus account to compose one.
So: the mesh-broker seat delivers mesh-bus; a module requires it and gets an
address, a sealed credential and the trust to verify the server; a module
that requires nothing has no account at all. The requirement delivers the
connection, the declarations shape the authority, and declaring a subject
without requiring the bus is refused as incoherent.
mesh-bus and nats are deliberately two names: a module may run its own NATS
as a backing service exactly as one provides amqp, and a manifest saying
"nats" would otherwise mean either the mesh's nervous system or a private
queue.
The seat's Delivers was wrong twice today — amqp, then empty — and the
comment says so rather than reading as though it were always right.
0117 went a step further than it had grounds for. It was right that the bus
is the only bus, and wrong that the amqp interface must therefore retire —
because it conflated two reasons to want a broker. Using one to reach
another module is a second bus and stays refused. Needing an AMQP broker as
a backing service, the way something needs a database, is ordinary, and
forbidding it would make the mesh unable to run normal software while
calling that architecture.
So the broker becomes a plain provider module: no seat, not foundation,
never raised at genesis, no retirement condition. lavinmq now claims nothing
and provides amqp; nats claims mesh-broker and provides nothing.
The rule that survives is about direction, not software: inter-module
communication goes over the bus. A module may hold a broker for itself; it
may not use one as a channel to another module. That is a review judgement
where 0117 could have used a parser, which is the honest cost.
0106's progressive insight was itself wrong and is corrected by a second one
there — nothing moves off the old broker, so its "one purpose" sentence does
not become true, it is just not what that server is.
The insight check needed two fixes it found itself: a date may carry
trailing words, and a bold run with a link is discussing an insight rather
than marking one. All four bad shapes still fire.
1.1 to 1.4 built and tested. 1.5 turned out to need no controller change:
it already resolves the broker by seat and names no broker module in its
source, which is what ADR 0079 was for. The genesis module set naming is
scenario and installer config, carried with the bed.
Recorded what must NOT change yet: the amqps:// credential shape and the
5671 default are correct until the rollout, because steps 1-4 leave every
node on AMQP.
29 put a module's events and tools in one namespace, 25 kept mesh.events.*
and mesh.tools.*. One namespace is right — a module's authority over its own
name becomes a single pattern the server enforces — but it needs a kind
token, because a stream is a subject filter and mesh.mod.*.> would persist
every tool call in the mesh. Tools stay on core NATS for the reason 25
already gives.
So: mesh.mod.<module>.event.<name>, .tool.<name>, and seats the same shape.
Found composing the first real configuration. Scoping every inbox to its
owner is right and leaves a responder unable to reply, because the answer
goes to the caller's inbox. The fix is not a wider grant but the server's
own allow_responses: one reply to the subject of a message the user actually
received. Without it every tool call times out while the permission list
looks correct.
Versioning: additive is free; a breaking change is refused while callers are
bound, and the refusal names them, because the mesh already holds the uses
graph; a real break versions the subject, not the seat name, so the role
does not fork; binding is a recorded pin, not a drift to whatever is newest.
Semantic change stays open — no fingerprint sees it, and saying so beats
implying the check is complete.
Provisioning: a provisioner's create/remove/holds IS a serves protocol, so a
provision interface is a seat that also delivers a credential — which is why
design 26 already allowed that. The per-consumer resource is what stops the
two collapsing into one.
Secrets: sealed, so the bus is never trusted with plaintext — but sealed is
not enough, because a stream persists and a durable ciphertext is an archive
the day a key leaks. So a secret never enters a stream: core request/reply
only, and a declaration names a secret rather than carrying one, which is
0098's fetch-don't-store applied where carrying is worst. The vault's own
credential and the bus's own accounts are the two bootstrap exceptions,
resolved the way 0067 resolves the control plane.
Also rewrote the addresses paragraph, which was too compressed to follow:
on-bus addresses disappear because nothing stores them, off-bus ones are
untouched and still 0098's problem, and the bus's own address is the one
that cannot be a subject.
The architecture 0117 opened needs a module to offer a service as a role on
the bus — one holder, addressed by what it does. A closed table in the
controller cannot express that: a capability a module contributes would
require changing the mesh itself.
But 0110 closed the set for a good reason — nothing could say what seats a
mesh had, and the hand count came out at eleven of thirteen. That argues for
enumerable, not hardcoded, and 0110 weighed free-form against a fixed table
without considering a third option: closed at any moment and derived from
the catalogue. A derived list cannot drift, which is how the count broke.
So: the mesh's seats stay the mesh's, reserved by the mesh- prefix so the
prefix is the rule and there is no list to maintain; ten seats are renamed
to restore 0079's convention; everything 0110 decided about what a seat IS
survives untouched.
Design 29 carries the declaration model: three namespaces, subjects derived
from local names so a manifest survives the wire changing, queues never
declared, five relationships (the job and state shapes 0041 had no room
for), and the build-publish-deploy lifecycle with hard, soft and build-time
dependencies distinguished.
0041 gets a progressive insight: "no per-consumer setup, only a
subscription" was a fact about a topic exchange, and a JetStream durable
consumer is a real object someone creates.
WBS 1.3/1.4 were wrong and say so: streams come at registration and
consumers at assignment, so only the foundation set belongs at genesis.
Issue 115 is resolved and converted four modules away from named volumes;
the bus's own data is not the place to bring one back. Also: NATS carries
TLS on the client port rather than beside a plaintext one, so there is no
5671/5672 pair to mirror.
A module does declare requirements the provisioner fulfils — but the broker
it gets that way is a private vhost, the analog of a database, not the
mesh's bus. Two modules of the new mesh depend on it, so the compatibility
broker was never single-purpose and its retirement would have stranded them.
NATS is the heart: one bus, a module's messaging is subjects on it scoped by
what it declares, and no module is handed a server of its own. The seat
delivers nothing; the interface retires with the broker. Also closes the
EVENTS question — one stream, on the bootstrap argument, not preference.
Designs 25 and 28 go in-progress: step 1 is starting.
The insight check caught a false positive on its own first real use — its
bold-run pattern crossed newlines and joined an unrelated `**` to the
marker. Constrained to one line, still catching all four bad shapes.
Both repository checks fail on main. 0115 is cited by no design, and it is
marked accepted while resting on 0112, which is proposed.
Design 27 already states the rule the record decides — "a module is assigned
at most once to a node, and that pair is the assignment's identity" — so it
is the home, and now says so. And 0112, 0113, 0114 and design 27 are all
proposed: the batch is under review, so the record is too. Promoting 0112
instead would be marking a record accepted to satisfy a check, which
check_rests_on names as a failure this repository already made once.
PR #133 landed a different 0115 while this branch was open. The bus record
is now 0116, with every citation in designs 19, 25, 28 and the index
following it.
Note: cycle.py and records.py both fail on main as merged, on that record —
nothing cites it, and it rests on 0112, which is still proposed. Both
pre-date this branch and are left for their own change.
A record can assert a fact that goes stale while the decision it supports
stays right. Superseding for that buries a sound record under a second one
and makes every reader work out which is live. So a correction of fact is
now made in place, marked and dated, with the old wording quoted — bounded
by three conditions and checked by records.py, which fires on an unmarked,
undated or back-dated note. Judgements still supersede.
Applied to 0115: no conformance suite exists to recapture, and the full
genesis bed cannot run until the links exist. Designs 25 and 28 follow.
Counting the surface first changed the plan twice: the genesis bed cannot run
until the links exist, so it belongs to step 4, and there is no conformance
suite to recapture — step 3 builds one against the current bus before moving
it. Both corrections are recorded in the breakdown rather than edited into
ADR 0115. Also indexes design 25, which was never listed.
The NATS change was recorded as one undivided item, which hid three gaps:
a mesh already running had no adoption path, the protocol specification did
not know its transport was being replaced, and nothing was runnable until
everything was. Dividing it is what surfaced them.
125: a hold is not a line in the apply report — sixteen resources held
for an untaken module while four surfaces reported success, and the
operator stopped the edge's predecessor on their word (the route-proxy
flip outage). 126: a changed volume path neither recreates a running
container nor warns, and a roll-out upgrade policy makes a build a
deployment — together they turned a data-path migration into a forge
outage (the /var/lib move). Filed as 119/121 in the novox session
before syncing; renumbered past the other session's 119-124.
Not every host path names this machine. A system file the mesh owns is at that path on every machine
of the kind — the path is the fact. The operator's shared data is already answered as an access. A
path inside a container is the software's contract. What is left, and what a node's default layout
would replace, is 514: a module's own data, and what the mesh writes for that module.
Documents the reservation model as the records already have it — a root per node, one directory per
assignment, a placement for adopted data — and the three things nothing states: where the root comes
from, what sits beneath it, and the order the 514 are retired in.
Stops there deliberately. Changing where a definition looks without moving the data does not fail: the
mesh creates the directory, the container starts, the service comes up empty. Retiring these is a data
migration with a verification step, module by module, and belongs with whoever can see the machine.
A path inside a container is not a fact about the machine — /run/secrets and the directory a server
keeps its data in are the software's own contract, true in any mesh that runs it. Only the host side
of a mount names where it landed.
The first sweep matched path-shaped strings, so it counted both halves of every mount and every
in-container location a value mentioned: 798. Counted by role — directory and file resources, the
host side of mounts, accesses, and the targets of binds, grants, receives and secrets — it is 698
across 70 definitions.
The five modules this report first named were what a first look found. A sweep of all 71: 49 public
domains across 26, the node's own name 58 times across 19, a routable IP 12 times in one, 798
absolute paths across 70 (issue 119's number, grown), and no email addresses at all.
The sharpest case is not a domain: a mail module states the node's public IPv4 as the address it
trusts a real-IP header from, so a node that moves or gains a second address stops attributing mail
correctly, silently. Two upstream resolvers are excluded deliberately — naming a public DNS service
is a policy default, true of any mesh, not a fact about this one.
The object store derives each consumer's bucket from the login the mesh minted, and never reads the
one a definition named. The consumer still has to tell its own software which bucket to use, and has
no way to be told: bound values come from the provider's serves, which is a literal block identical
for every consumer, and a provisioner returns nothing. So all three consumers wrote the answer down
by hand and one of them wrote the predecessor's bucket — a key scoped to one bucket and software
asking for another, which reads like a credential fault and is not one.
Records the general shape: any interface where the provider names the resource forces the consumer to
reproduce the provider's rule, kept in agreement by hand and checked by nothing.
Three wordings disagree, and the confusion is the damage: the glossary defines artifact as an OCI
image while the build vocabulary already names four kinds in use, two of which are not images; the
image registry's seat is named after its job while ADR 0079 names foundation seats after their servers
and ADR 0109 names package seats after their ecosystem; and prose that says 'the module's image' reads
as though a module were an image.
Records the question the naming hides: ADR 0075 keeps two provisions because packages and images are
two protocols, and already allows the forge to provide the artifact store. The second implementation
rests on a bootstrap argument, and the forge has the same upstream-server shape the store and broker
have, which ADR 0078 raises as plumbing and adopts in place.
Step 4 claimed the forge cannot exist as a container before the builder has built its image. The
forge's server is an upstream public image pinned by digest; only its runtime sidecar is built. The
store and the broker have the same shape, and ADR 0078 raises both at genesis as plumbing and adopts
them in place — so the forge can be raised the same way and serve git, packages and OCI before
anything is built.
What survives: a grant is minted by the provider's runtime sidecar, which is built, so the question
is whether raise-service, grant, build-sidecar simply works. Sequencing inside the mesh, not images.
The wrong version came from taking a record's bootstrap argument at face value instead of comparing it
to how the store and broker are raised — one command away in the manifests.
Two things were treated as one. The mesh's own artifact store holds the store seat and is internal by
design — reached by name over the overlay, no accounts, ADR 0082. Serving a registry publicly is a
service the mesh can host: a module with its own name, accounts and storage, like anything else it
runs for somebody. The conversion this report was written beside gave the seat holder a second public
door over the same filesystem, which is neither.
Keeps the original issue whole — no garbage collection, and the settings a routine needs are not
enabled — drops the two-door complication, and sharpens one thing: deletion on the only door is
deletion on a door with no accounts, which the predecessor kept behind its authenticated one.
Asked whether merging the three reviewed changes would set a precedent. It would not: five modules
already carry a name belonging to this one mesh — a workflow module stating its host, protocol and
absolute webhook URL, and two carrying a full clone URL for a repository on the mesh's own forge.
That changes what the issue is for. There is no version of this catalogue today that does not name
the mesh it was written in, so refusing three changes buys nothing and a mechanism is the only thing
that removes any of them. The three were merged on that reading, each PR saying so.
Three open module changes independently wrote one mesh's names into the catalogue — two literal
public URLs, because the software generates absolute URLs behind a proxy, and one bucket renamed to
match what exists here. None was careless: the mesh composes <label>.<public-domain> for the proxy
and never hands it back to the module that asked for the route, and no interpolation yields a public
name, so writing the answer down is the only expressible option.
Files it rather than blocking the three, because the fix is a mechanism and the instances are live
needs. The cost is stated: a second mesh installing the identity provider gets the first mesh's
hostname, and nothing distinguishes a literal domain from a version number.
Both halves are on their main branches, so the seats stop being an intention. Writes the as-is
document from the controller's code and the catalogue's manifests: the closed set of fourteen, the
three refusals a claim meets, the holder being an assignment and nothing else, and the one place a
seat changes resolution — which of several providers answers, never whether a requirement may go
unanswered.
Two things the as-is layer exists for are stated rather than smoothed over: a seat cannot answer
before it is held, which is the standing condition issue 121 records; and capacity is not
implemented at all, so the design's bench has no counterpart in the code.
Asked first whether the seats change fixed this in passing, since it landed the same day and
touches both manifests the report names. It did not: the seat's holder is consulted only where
several nodes provide the thing, and with none providing it resolution refuses outright. Genesis
has no exemption — the unchecked first pass exists to learn what each node offers, and a
declaration is never built from it.
Records the part that did change: the three tests left failing on purpose were deleted by the
controller's seats PR and replaced with passing seat-based ones, so the gap is invisible again.
Adds the resolver to located-in, since that is where the refusal is.
The seats half of to-be 27's review is settled, so the two records it rests on are accepted and
the vocabulary catches up: the glossary's *seat* becomes a named role from a closed set, held by
an assignment and possibly delivering a provision, and 23 — Choosing a provider gains the seat
step in resolution, with ambiguity still refused rather than guessed. Both were held back when
0110 was proposed, because a document may not rest on a record that is not accepted.
26 — The seats moves to in-progress rather than designed: it names the files that implement it,
and naming a file claims implementation, which is only defensible once those files are on the
owning repositories' main branches. It becomes implemented when mesh-controller #63 and
mesh-catalog #69 land.
0112, 0113 and 0114 stay proposed; to-be 27 stays proposed with them.
Renumbered from 117, which is taken on main by 'a module's own code is a container in one record
and a process in another' — two reports claimed the same number and git would not have said so.
Scrubbed the node's name and a real registry path; this repository is public.
Keeps 118: the other claimant to this number is on main as issue 119, where ADR 0112 points.
Scrubbed the service's public name — this repository is public — and completed the report's
frontmatter with the fixed-by and amended-design keys every other report carries.
The operator's wish written as intended behaviour for the NATS bus:
every loop compares against what is, repairs by the ordinary path, never
destroys, and raises a condition for what it cannot fix. What is done
before NATS is limited to what survives the move.
The harness compares against its own memory, so a backend that loses
what was provisioned (the cache's ACL users on a server restart) is
never provisioned again, silently.
The fact-check found mailu, whose user is its mailbox, so 0114 rotates
over two credentials rather than two logins, the adapter choosing what a
credential is. Also: minio keeps non-empty buckets; five backends take
their admin credential only at first init, so single-party rotation is
staged; postgres ownership moves to a non-login role; the harness keys by
consumer; rotation state lives with the vault. Consistency fixes across
0110-0113, 26 and 27; issue 103 resolved by mesh-host PR #22.
Graduates research 016. Retiring a login is separated from removing a
consumer, which closes a data-loss path in five providers; single-party
secrets rotate in place; the number of parties decides, not the provider.
Overlap as drafted in 0113 would have deleted consumer data: seven of
eight providers name the resource after the login and five drop it on
remove. Rotation is now undecided in 0113 and to-be 27, pending the
survey. Also: a requirement naming a seat resolves to its holder, a
person chooses among remaining candidates at assignment, the controller's
secrets are requirements of its definition, genesis seals to the
control-node key, and moving the vault or broker is break-glass.
Decided with the author. A credential is never changed in place: each consumer has two logins, both
derived by the mesh, and uses one at a time. An applier adds the new login beside the old through the
adapter's existing create, and confirms both work; only then are readers released to the new one and
restarted by derivation; only when every reader has confirmed is the old login retired through the
existing remove.
It closes the three cases review found in applier-first rotation: an offline reader keeps working on
the old login until it returns; a bus account's owner keeps its bus until it has moved; a provisioner
restarted mid-rotation is still delivered both values. Nobody is ever without a credential that works,
which replaces to-be 13's all-or-nothing rule with a stronger one.
No consumer module changes. The alternation is the provider loop's. A provider's adapter gains one duty,
giving both logins the same rights over the consumer's data — in postgres, membership of one role that
owns it. The mesh derives two logins per consumer, both within ADR 0049's limit, which 0113 now names
among what it amends. Every rule has a check: overlap, offline reader, bus account, restarted
provisioner, equal rights, login length, and confirmation only once the old login is gone.
Decided with the author:
- A seat is held by one assignment, not claimed by a definition. A definition says which seats a module
can hold; an assignment says which it does. The store module can run on every node and one assignment
holds mesh-store; moving a role changes an assignment, never a definition. The foundation's seats name
what the mesh itself uses and route no consumer — database and amqp consumers use co-location, the
holder included. This replaces the wrong rationale that the foundation's store is "provider to nobody",
which contradicted ADR 0078 and to-be 21. 0079's one-postgres rule becomes one mesh-store holder.
- A module is assigned at most once to a node. The instance identity in 0112 and 27 is withdrawn, and the
login-length problem with it.
Review fixes to 0113:
- The bottom of the stack: the vault is installed as soon as the shared runtime base exists, and genesis
generates everything needed until then — including the permanent controller's, the control-node
agent's, the builder's and the broker provisioner's bus accounts, and the controller's store login.
Genesis creates those accounts until the broker's provisioner runs and adopts them.
- Genesis's values are delivered recorded as the mesh's own, so 0092's never-replace rule for operator
values does not make them unrotatable.
- Backend-issued secrets (a forge's once-only API token) enter through the vault. Non-module parties
(the controller's logins, node agents' accounts) are answered the same way, the controller asking on
their behalf; an enrolment token reaches the controller only as what verifies it.
- A secret with no provisioner to apply it is marked not rotatable by the mesh and refused, instead of
a restart reported as done. Unused password generators in six provider clients are removed, and a
catalogue scan checks no module mints.
- Rotation's lock-out cases (offline reader, bus account owner, restarted provisioner) are recorded as
open, with overlap and re-confirm-with-safeguards as the two answers, to be chosen before acceptance.
0110, 0111 and 26 are marked proposed: they changed in meaning and are under review, and an accepted
record must not rest on proposed ones. To-be 23 and the glossary are restored to main; they change when
these records are accepted.
Two decisions taken with the author:
- Genesis delivers and the vault adopts. The vault cannot run first — it is built on the runtime base
the installation makes after the store, broker and controller, and it learns its work over the bus.
Genesis generates the foundation's first shared secrets, seals them to the operator key, and
delivers them to the vault through the path an operator's value takes; from then on the vault holds
and rotates them. This answers ADR 0085's own reason for rejecting vault-only minting, which 0113
now names instead of stepping around.
- Rotation re-confirms on every pass. An applier repeats its confirmation until acknowledged, so a lost
message costs one pass; an applier that stops after applying locks readers out until its supervised
restart, and that window is stated and shown, not claimed away.
Fixes:
- Scope: a shared secret is made by the vault; a private key (node sealing keys, the operator's key,
the certificate authority) is made where it is used. The inventory adds the makers the first version
missed: node and builder broker passwords, and enrolment tokens.
- Broker accounts are created by the broker's provisioner, not the controller, so the controller never
holds their plaintext; mesh-broker delivers amqp again — one broker per mesh — and only mesh-store
delivers nothing.
- secret is a reserved provision: only the mesh-vault holder may provide it, and no pin routes around it.
- A secret's contract says whether a recipient applies it or reads it at start; appliers are never
restarted for it, init-only secrets are applied, and confirmation is to-be 13's standard.
- Operator secrets are one rule everywhere: a secret requirement answered by the vault (0112 no longer
says otherwise). A data provider's adapter may return fields; the data-return check names a lab consumer.
- 'Holder' now means a seat's holder only; a secret has recipients.
A secret comes into being seven ways today: provider credentials, own secrets (54 modules), broker
accounts through a command that is easy to forget, a vault that only records what the controller
mints (6 modules), operator values, licences, and root secrets. The vault was built to end own secrets
and did not; the old path was never retired.
0113 is rewritten as a waterfall. The vault makes every secret and nothing else does. A provider that
needs a secret for a consumer requires it from the vault, declared once in its provision's contract
and expanded per consumer by resolution; the vault delivers it to both holders, each sealed to its own
node, so a provider's code is unchanged. Own secrets, broker passwords, operator values and licence
credentials take the same path. Genesis is not an exception: it raises the vault first and asks it,
so there is one way a secret is made from the first one on. The vault can sit at the bottom because it
requires nothing but a broker account.
One shared mint function in the SDK was considered and rejected: generation becomes uniform but custody
stays spread over every provider's machine, and each SDK language needs its own implementation.
Rotation is asked of the vault and is provider-first: the value goes to the holder that accepts it,
which confirms, before the holder that presents it gets it, so the lockout window shrinks to the
consumer's own restart, and an unconfirmed provider holds the rotation rather than half-doing it. The
host derives which processes to restart or recreate from the requirement a definition reads, so no
definition declares restart-on for a secret. A rotation shows unconfirmed until each consumer restarted
and passed its health check. Issue 103 becomes a prerequisite.
The file is renamed to match what it now decides. 0112 follows.
0113 — the plaintext claim was false under its own mechanism: handing a provider's answer to the
controller puts every secret on the broker and in the controller in the clear. The provider now seals
each secret field itself, to the consumer node's public key the mesh hands it, and the controller
carries sealed fields it cannot open. That is stricter than today, where the controller holds every
minted credential in the clear. Option 3 (plaintext to the controller) is recorded and rejected. The
foundation exception now covers root-secret rotation (0085) and forms like the broker admin's hash, so
no phase claims to remove the broker's bootstrap step. To-be 24 and 13 are named among what it amends.
27 — resolution is consistent with 0110: co-location and the only provider apply only where no seat
delivers the provision, so an unheld seat is refused even with one provider. The secret-field rule now
matches 0086 exactly (a declared env-file, never a container environment value). The seat placeholder
is the controller's, and the one module reading it moves to a host port. Contracts are held by the
controller and written down in phase 1, so they can be checked; every rule has a check. An operator's
secret is still the operator's, with the vault as custodian. Which seats a module holds is listed as
not settled.
0110 — the unheld-seat-with-one-provider case and the one-answer-for-everyone rule have checks; the
claim about moved manifests is corrected. 26 — the table governs and the code catches up, not the
reverse; scope and capacity agree with the glossary; moving a seat is described as it really is today.
0112 — aligned with 27, and lists 0049 and 26 among what it changes.
Issue 118 is renumbered 119: another branch took 118 first. 'Control-plane' is gone from 0110 and 0111.
The design pass. Everything a module needs is a requirement: a name, a contract, and one of four kinds
of provider — a module, the node's host, the mesh, the operator. Installing a module resolves every
requirement or refuses, naming everything missing at once. It retires six mechanisms that grew
separately: provisions through bindings, settings, assigned ports, machine facts, minted secrets and
literals in the definition.
ADR 0113, proposed: a provider makes what it provides, and the mesh carries it back sealed to the
consumer's node. It is the return path ADR 0048 left "to a separate decision", now needed three ways:
data provisions with nothing to answer with, contracts needing a value the controller cannot make, and
a vault that generates nothing. Who a consumer is stays the mesh's (ADR 0049). Genesis is the one
exception. On acceptance it supersedes 0048 and amends 0085.
ADR 0112 is revised from three sources to that single concept.
ADR 0110 is amended for two points raised in review. The vault gets the mesh-vault seat (issue 106).
A seat's holder outranks co-location for a provision it delivers. Writing that down exposed an
inconsistency: mesh-store delivering postgres-database would have sent every database consumer to the
control-node, against to-be 23's node-local stores. So a seat delivers a provision only where the mesh
has one answer for everyone — artifact store, npm registry, git, vault — and mesh-store and mesh-broker
deliver nothing. 23 and 26 follow.
'Control plane' becomes 'controller' in the records written today.
- Secrets follow ADR 0085 as amended: a module's own secret is a provision the controller mints and
the vault records. The previous commit had that backwards. Whether the vault should generate
instead is recorded as an open question, not decided.
- A directory's contract is owner and mode only. The persistence flag was the keep flag ADR 0030
refused; a directory is kept while it holds anything, and disposable data is a named volume (0107).
- An operator's shared data stays an access (ADR 0051), which rejected an operator-owned directory.
Only where its path is written moves to the assignment.
- The records it changes on acceptance are named: 0051, 0091, 0046 (settings keyed by instance),
0084 (a provider is a node and an instance), and the glossary, which gains its new words only
when the record is accepted.
- How it is checked covers every stated rule. Container-side paths are no longer flagged by the
host-path rule, and code fallbacks are covered.
- Provisions are what other modules provide. A seat's occupant is not listed as one, and the vault
is not described as selectable per assignment.
- 'Control plane' becomes 'controller'. The provider count is ten of eleven, not eleven of twelve.
The first draft listed minted secrets under what the mesh generates. ADR 0085 made a module's own
secret — a password, an internal token, an external key it was handed — a secret provision answered
by the vault, like a database by the store. What the mesh still mints is the delivery credential for
each provision a module takes (ADR 0048), the vault's own included.
Issue 118 records what a review of where module code reads its files found: 789 host-path strings
in 70 of the catalogue's 71 definitions, every one a decision the definition makes about a machine.
Mounts are checked (ADR 0091); the same paths retyped as values are not. It records what that has
already allowed — a DNS provider that would provision nobody silently, a contributions file that
names credentials by host path and so forces every provider to mount at the identical path, an SDK
loop that treats an unwritten contributions file as empty without a word, defaults in code that
disagree with their own manifests — and that no module can be assigned to one node twice, because
every identity is keyed by the module's name.
ADR 0112, proposed for review, answers it the way ADR 0038 answered ports: a definition names
variables, and installing it resolves every one or refuses, from three sources — the assignment's
own configuration, provisions the mesh resolves against a contract, and what the mesh generates or
knows. A directory becomes a provision: the module requires one by name with its owner, mode and
persistence, and where it lands is the assignment's. The mesh's own files stop carrying host paths.
An assignment gets an identity of its own, so a module may run twice on one node.
Checking copies for agreement was rejected as checking something that should not exist; rewriting
paths per assignment was rejected as inferring which strings are paths by their shape. Syntax, a
node's default layout, and when a second instance becomes possible are left to the design.
The enumeration behind the first set read manifests in two repositories and missed a claim made
in the control plane's own code: the private-network module it ships claims the-private-network at
node scope. A closed set without it would refuse the control plane's own module. Thirteen claims in
use, naming twelve seats.
Seats have been doing two jobs and neither is written down. The mechanism ADR 0009 introduced is
enforced — a second holder is refused — but any well-formed name becomes a seat by being claimed,
and nothing can say which seats a mesh has or who holds them: holdings are assembled while planning
and discarded. The enumeration done while preparing this missed the control plane's own manifest,
because core modules' manifests live in its repository rather than the catalogue.
0110 closes the set. Each seat has a name, a scope, what occupying it delivers, and the record that
made it one; a claim outside the set is refused. A seat is held by a module assignment, and what the
mesh knows about the holder is what it knows about that assignment — nothing is stored beside it. A
seat may deliver a provision, and then its holder answers for it among several providers: pin, then
the holder, then the only provider, then refused. That keeps 0009's "refused, never guessed": the
seat is the choice made once, mesh-wide, instead of a pin per consumer node. The first set is the
eleven seats already claimed plus 0109's npm-package-registry, so nothing in use is refused.
Two concepts — seats for exclusion, a new word for consumable singulars — was rejected: both mean
"this mesh's one X", and the overview a person wants is one list.
0111 gives the mesh a git seat and makes a build source one of two explicit forms: a repository on
the seat's holder, recorded by its path and cloned from wherever the holder runs at build time; or
an external URL, recorded and cloned exactly as given. Recognising self-hosted sources by matching
URLs against the forge's address was rejected — it fails in the one case it exists for, after the
forge moves. Credentials for private repositories are left undecided and said so.
Design: new to-be 26 (the seats); 23 gains the seat step in resolution; 18's source entry names
the two forms; the glossary's seat and provision entries say where they meet. 0109 is carried from
its own branch so every link here resolves.
Extends ADR 0075. Surfaced fixing builder's hand-faked package-registry
binding tonight: gitea's manifest declares the provision once with a single
npm-path, conflating what should be independently assignable per ecosystem
(npm/cargo/docker/...) the same way artifact-store and package-registry
were themselves split. Cited in 22-the-work-ahead.md's Phase 2, where the
target state this decision points at was already described a week ago.
Numbered 109, not 108: route-proxy's policy feature (mesh-controller PR
still-unwritten decision record — reserved but never committed. Renumbered
around it rather than colliding.
Fixing builder's hand-faked package-registry binding tonight (requires:
package-registry, a real mesh grant instead of a hardcoded JSON fragment)
broke three tests describing a deliberate carried-binding fallback for
exactly this: gitea's own image is built by builder, so builder cannot
yet hold a real grant from gitea the first time either has to exist.
Invisible on novox (already bootstrapped, gitea already live) — real on
any genesis from scratch. Fix left in place, tests left failing rather
than reverted or hacked, so the gap stays visible.
Filed 2026-09-24 on a branch of its own and never merged, numbered 113, which is taken. 114 is
free because a sibling branch folded it, so it takes that number and keeps its commit.
Kept separate from issue 117 rather than folded into it. 117 asks the same question of every
module and locates the missing decision; this asks it of the controller, where `network: host`
means container network isolation — the property that resource type usually buys — is not in use.
That observation is this report's own and is nowhere in 117, and folding would lose it.
Its first open question is answered by 117's diagnosis and now says so: the host's `process` shape
is built, applied and tested, restart and run-to-completion semantics included, so deciding this
does not wait on host-side work.
Filed after a session where every mesh-controller interaction went through
docker exec — its manifest runs it as a container with network: host, using
none of the isolation that resource type usually buys, while ADR 0006 makes
it the mesh's single point of coordination. Open question, not a claimed
defect: does type: container get the controller anything type: process
(supervised the way the host supervises its own unit, per ADR 0005) would not.
Asked what the "sidecar" is and whether a supervised process would do instead. The repository
answers both ways. ADR 0047 (accepted, unsuperseded) says a module with tools or events runs a
container carrying its compiled code. To-be 18 and 20 (both proposed) define a `process` resource
type — the module's own code, a unit the machine's supervisor keeps up — and the worked guide says
plainly "it is why these are `process` rather than four containers." Neither design doc names 0047,
and no decision record mentions a `process` shape at all.
Diagnosed rather than left open, because the ground truth settles what the report could not.
The shape is real: mesh-host defines TypeProcess, applies it, and tests it, and the host's
vocabulary is twelve shapes rather than the nine ADR 0029 counted. So the alternative the report
offered — that two proposed documents describe a type that does not exist — is disproven.
ADR 0029's mechanism is intact and was not enough. The vocabulary-count test names the decision
behind each addition: network 0029, access 0051, opening 0100. The eleventh names a *proposed
design document*, and TypeProcess is the only shape in the vocabulary whose doc comment cites no
ADR. Requiring every addition to name something does not require it to name a decision.
The argument this issue asked for already exists — as a Go test comment. "It is a full-host shape
rather than a portable one: it needs a process supervisor to install into. It does NOT need a
container runtime, which is the point — only software that genuinely needs isolation asks for a
container." That is a decision's context and consequences, in another repository.
What the catalogue does is a third thing: 115 container declarations against 3 process, all three
in showcase — the module to-be 20 documents. There the tools resource is a container running
`sleep infinity` on a bare upstream base with the broker credential mounted, and the tools and
provisioner entrypoints are run by nothing. That is the condition 0047 was written to end, back
in a new shape.
Where the isolation argument leaks is narrower than expected and worth having precisely: the
serving key and the credential shape both conform. But serveTools serves every registered module
over one broker connection, the runtime takes its modules from a comma-separated list, and
x-source is stamped from the single credential — so two modules in one runtime means the second's
events are attributed to the first. Nothing refuses it and no test asserts against it.
Located on hq rather than on a code repository: the implementation and the design layer agree,
and the missing thing is the record. Which shape is right is left open, deliberately — this
establishes that the question was answered in practice and never written down, not which answer
is correct.
One correction kept in the trail: the first search here was for len(Vocabulary()), found nothing,
and was two steps from being written up as "the mechanism ADR 0029 relied on is gone." The test
binds the slice to a local first. A negative search result read as a fact about the world is the
same error issue 113 recorded.
The gap is closed in the proxy: policy applies, the four capabilities exist, the table is keyed
by host and path with a total ordering, and the two failure modes that rot quietly are held by
tests — a declaration carrying a credential refused rather than served, an unreadable secret
failing closed.
Resolved rather than left open because the issue reports a gap in the proxy and that gap is
gone. But the record says plainly what it does not yet allow: an operator still cannot move the
affected routes, because that needs the mesh side — a manifest able to declare these values and
the controller minting the secret auth names. Until both exist the capability is reachable only
by writing the routes file by hand. That is the ordinary build-out of a contract this issue's
decision created, and it belongs to to-be 08 rather than here.
The open questions are marked answered and kept rather than deleted, pointing at ADR 0108 —
what was rejected and why is the half worth having, and a section still saying "the fix should
not be written before these are answered" after the fix was written reads as though nobody
looked.
One finding kept in the record: priority was read with the reader for ports, which caps at
65535, and the one real rule this reproduces is declared at 100000. It parsed to zero, so
refusal and path scoping would have shipped looking complete and doing nothing on the only case
that motivated them. A validator borrowed from a neighbouring field is a silent default.
Issue 116 found the mesh's proxy applies nothing to a request — host lookup, forward. Against
what the replaced ingress actually relies on, four capabilities are missing: authentication
(three dependents, each gating an admin surface with no login of its own), refusal scoped to a
path (one, a live incident mitigation), path-scoped routing with priority, and redirect.
Policy goes on the route rather than beside it. A proxy-side settings layer keyed by route name
would keep the grant literally clean, but then "what protects this route" is answered from two
files nothing keeps in step — and a route's protection is part of what a route is.
The set is closed at those four, so a fifth is an amendment and each addition is earned by a
dependent that exists. An open middleware surface was rejected: it recreates what is being
replaced, and narrowing one later is far harder than widening a closed one.
Where policy needs a credential the declaration names a secret and never carries the value,
which keeps the existing secret machinery the only thing holding credentials. Inlining a hash
was rejected as the first credential in a declaration — a precedent easier to set than withdraw.
This re-keys the routing table by host and path with priority, which follows from the decision
rather than being a separate one: two of the four need one host routed more than one way. Equal
priorities must resolve identically every time or the proxy stops being reproducible.
The record says how it is checked, including the negative case that rots quietly — a
declaration carrying a credential value rather than a reference must be refused, so the
rejected option cannot return by accident.
08-connectivity §3 names the record and gains the subsection; issue 116 gains amended-design.
Two things the report got wrong, and one it could not have found the way it looked.
Disclosure first: it carried a real hostname and an absolute node path, in a public
repository. Both are gone; the ingress, the modules and the routes are named by role, as the
rest of 04-ISSUES does.
The count was low. Basic authentication has three dependents in the catalogue, not one — the
key-value store's browser UI, a database web UI, and the ingress's own dashboard. All three
are credential-less admin surfaces whose only gate is a middleware the mesh's proxy lacks.
The earlier version read only the node's dynamic configuration directory, which cannot see
what modules declare as container labels; counting needs both sources, and the report now
says so.
Two gaps were missing entirely. Redirect rules: two live routes canonicalise a www name onto
its apex, they exist only on the node and not in the catalogue, and they fail silently rather
than erroring. And path-scoped routing with priority, which is the one that reorders the
issue: the table maps host to exactly one target, so a host cannot be routed two ways, and
the refusal rule matches a path on a host already routed elsewhere. Authentication and a
source filter would not make it expressible. Path scoping is a prerequisite, not a sibling.
Also corrected: the refusal rule was described as an address-scoped deny. It is an allow-list
holding a single documentation-range address — deny-everyone — so reading it as address-scoped
points at the wrong fix. And its severity was understated: its own header records it as
incident response closing an abused write primitive, which is not "a real exposure" but a live
mitigation.
The open questions now say plainly that they are design questions and the fix should not be
written before they are answered, and one is added: whether a declaration may carry a
credential at all.
Comparing route-proxy against what HAL's actual Traefik config does today,
not Traefik's general feature set, per the standing rule that the nox mesh
must do at minimum what the HAL mesh it replaces already does. Everything
else checked out even or better; these two are real, confirmed gaps —
RedisInsight has no login of its own and depends entirely on Traefik's
basicauth middleware, and the gitea-internal route depends on an IP-scoped
deny rule. Neither has any equivalent in route-proxy's single-lookup
request path.
The object-store module was repinned to a maintained fork of the withdrawn server image, its
runtime sidecar built rather than pulled, and its data moved off the predecessor's live
directory. The instance is closed; the three general points the report makes are not, and What
was done says so rather than letting a resolved status imply otherwise.
Folds in the one thing a duplicate report of this symptom had that this one did not: issue 064
asked whether the build environment can reach a declared vendor image and assumed that, once
declared, it stays fetchable. Withdrawal is the case that assumption does not cover. The
duplicate is not merged — it carried the reading this report's diagnosis retracts.
The previous version ranked candidates on whether they preserved single sign-on to the
object store's console. That is not a requirement: a "user" of the store is normally an
application, so the requirement is per-application keys scoped to buckets — which the mesh
already mints. And the console login it ranked on never worked; the module's own hook comment
records "policy claim missing", a failing login written up as progress.
It also omitted the incumbent's own maintained fork, which changes the question from "which
product replaces it" into two decisions: repoint, or migrate — and if migrating, to which.
Repointing costs an image reference; migrating costs a data copy, two handler rewrites and a
maintenance window. Repointing does not foreclose migrating, which is the argument for taking
it first.
On the corrected requirement Garage ranks first — its per-key-per-bucket model is the
requirement verbatim, its admin API matches how the mesh provisions, and the highest-risk
consumer is first-party documented against it. Its remaining gap (no versioning, no
server-side encryption, partial lifecycle) is unmeasured against the buckets and is the one
thing that could still disqualify it.
Measured and folded in: 230 GiB logical, 82,496 objects, 468 GiB raw at 2.03x, eight drive
directories on one filesystem on one machine. That last fact decides more than any feature —
the erasure coding is not buying independent-drive redundancy, so the redundancy model is
close to irrelevant and only storage overhead remains, which at this volume is a rounding
error against the headroom.
Both errors are recorded at the end of 01 rather than quietly fixed. A configured feature is
not an observed one; and when a dependency dies, "who took it over" precedes "what replaces
it" — searching for alternatives by construction returns things that are not the incumbent.
The diagnosis carried a table headed "claims that could not be substantiated", denying a
module.json, a digest pin, and an all-zeros runtime digest. All three exist. The table is
withdrawn in full and replaced with what is actually true, plus the two claims that remain
genuinely unverified rather than disproven.
The cause: one repository was searched and absence in it was written up as absence. The
catalogue of the mesh being built is a separate repository, not checked out where the search
ran, and all four claims were about that repository. Compounding it, the predecessor's
object-store module and the one being cut over to were treated as one thing — they are
different files in different repositories, one pinning a tag with no sidecar, the other a
digest with two container resources.
Also corrected in the report: located-in named the wrong repository; the "pins a tag" passage
described the predecessor; the open question about pinning by digest is struck, because this
module already does and it made no difference — a deleted digest resolves to nothing either
way. The section on why nothing broke is now scoped explicitly to the predecessor's
machinery.
The lesson kept in the record: "zero occurrences anywhere in the tree" is only as strong as
the tree searched, and a diagnosis must say which tree. A confident rebuttal of a correct
report is worse than no diagnosis — it sends the next person to the wrong place with a
written record behind them.
A thought raised mid-task needed remembering but not working on, and there was no
mechanism for that — so it was recorded by hand. This is that, made repeatable.
Records to Claude's persistent memory rather than the repository, deliberately. A parked
thought has no number, owner or status: giving it one asserts triage that deferring says
has not happened. A shared "deferred" document would be a central status file, which
AGENTS.md forbids. And a repository write means a branch, a commit and an MR — the drift
the skill exists to prevent.
Wraps no playbook, because deferring precedes the development cycle rather than being part
of it. It does say which playbook a thought would need if it graduates, and that recording
"undetermined" is the honest answer when the evidence does not say.
The stop condition is the substance: at most two lines, then return to what was in
progress. No plan, no triage question, nothing opened.
Records the rule the operator gave directly, mid-session, after checking
that HAL's own postgres and lavinmq both used a directory bind and the
mesh's adoption of them three weeks ago switched to a named volume without
a reason recorded anywhere.
Already built and rolled out on novox (mesh-catalog PR #54) before this
record -- urgent enough to fix first and write down after. Includes the
incident: the new host directories needed the container's own UID, which a
named volume gets for free and a directory bind does not; mesh-store
crash-looped on Permission denied until ownership was matched to what the
original volume already had.
Closes issue 115. Checks pass.
SeaweedFS was scoped as primary because it looked like the only candidate preserving
OIDC console login. Measured: its admin UI is Apache-2.0 but its identity-provider
integration is not — console SSO sits behind the per-TB commercial licence, alongside
point-in-time recovery and automatic EC repair. The free build gives OIDC on the S3 API
via STS and a console authenticated by local username and password.
So the answer to the gating question is that no candidate preserves the current feature
set for free, which this effort had written down as a possible outcome. Reopened across
three candidates with the requirement-by-requirement evidence in 01.
Two corrections to what the overview recorded. RustFS is not a binary-level drop-in
retaining existing data: API and on-disk compatibility are separate paths and the on-disk
one is preview-scoped. And it carries an open defect in the credential path the bucket
provision depends on, which gates it specifically.
Nothing graduates before two measurements named in 01: whether an authenticating proxy
is an acceptable answer to console SSO, and which S3 endpoints consumers actually call —
the latter because Garage does not implement the full span and cannot be ranked until
that is counted.
The decision (the mesh assigns a container's machine-side port; a module
says only what it needs) was proposed 2026-09-01, and the machinery
already implements it in full -- internal/inventory/ports.go's PortFor,
declaration.go's publishedOn. What was missing was the catalogue actually
complying: 14 of 46 modules baked a machine-side number into their own
manifest anyway. mesh-catalog PR fixes 11 of them (the two defensible
kinds -- foundation, protocol-fixed -- are left alone, per the issue's own
categories). Accepting the decision now that it's actually enforced, and
closing the issue it was blocking.
Checks pass.
Review of my own text found an unattributed claim — "the mesh's claim that a node
can be rebuilt from its declarations" — which is not a stated principle anywhere.
Replaced with the design position that genuinely covers it: to-be 07 chooses
references over payload because "reproducibility comes from pinning the identity of
a thing rather than carrying its bytes". This incident is that choice's failure mode
when the identity stops resolving, which is a sharper point than the one I made.
Scope stated honestly: the passage is about the foundation bundle and this module is
not in it, but pin-identity-fetch-bytes is how every module gets third-party images.
Also names the tension the mirroring question actually carries — mirroring is a move
away from references-over-payload, so it is a decision, not a fix.
The symptom arrived diagnosed as "the registry disabled anonymous pulls for the
whole vendor namespace". It did not hold: sibling repositories in that namespace
pull normally, the "$disabled" token field appears on every repository including
working ones and describes signing rather than access, and "actions": [] with a
401 is byte-identical to what an invented repository name returns. Both registries'
own APIs establish deletion instead.
Recorded because the correction is the expensive part to rediscover, and because
the instance was harmless while the standing condition is not: no node that does
not already hold the images can ever provision the module again, and nothing
detects that until one tries.
Research 015 scopes the replacement. It is not a redesign — the foundation design
already commits to S3 the protocol rather than the product, and the object store
is an ordinary module, so this instantiates an existing principle. The live OIDC
wiring is the requirement that gates the choice, and it is checked first.
Playbook 03 step 2: move status to diagnosing, then located once the
owner is known. located-in is filled with four confirmed packages —
the owner is known.
The carried-peer record (CarriedPeer/TunnelPeer) lives in mesh-controller
internal/inventory, not internal/catalogue. internal/catalogue is the
right package for the zone-generation side of the fix (facts.go's
nodeZones), but a different concern from where the name field itself
would go. Split the two so a decision record doesn't get pointed at the
wrong package.
Fixes, each named where it was wrong:
- reload-on is a service field; a container only has restart-on, which
recreates. Cited precedent (registry-trust-reload) is a service resource,
not a container. Fix: nats-server's own SIGHUP reload, triggered by an
in-image entrypoint watching a directory-mounted config file (issue 103's
recreate-on-change applies to a directly-mounted file, not a directory's
contents) -- asks nothing new of the host.
- A JetStream-delivered message's Reply field is already claimed by the
consumer's own ack address, so a responder using it answers nobody. Fix:
every CONTROL message needing a reply carries its reply subject in its own
payload; the controller publishes there explicitly, never via Respond().
Enrolment is the case this design actually depends on, so it's fixed there
too, not just noted.
- The listed permissions never granted publish on a durable consumer's own
ack-reply subject -- a module could receive but never ack, so every
message redelivers forever. Fixed with a scoped grant per module's own
consumer.
- One account (a deliberate choice, kept) means inbox privacy is the
permission list or nothing. The design granted 'its reply inbox' without
scoping it, which read as any user reaching any inbox. Fixed: each user's
inbox prefix is derived from its own identity and its permissions name
only that prefix.
New open question from this revision, not closed: whether the in-image
watch-and-SIGHUP shape belongs in mesh-sdk if a second module ever needs it.
Checks pass (records.py, cycle.py, index.py).
Checked why ADR 0104's forward-to-predecessor shape doesn't transfer to the
resolver the way it did the proxy: DNS is one process on one port, and
assigning the mesh's dnsmasq module replaces it in place, so there is no
predecessor process left standing to forward to.
But /etc/dnsmasq.d/hal-dns.conf's static address= lines for ace/shanks/g14
match the mesh's own carried-peer addresses from overlay show exactly. The
name a carried peer needs isn't a guess the operator has to make under
pressure — it's a transcription of a record the predecessor already has and
has been correctly serving for six days. Located in mesh-controller (no way
to attach a name to a carried peer today) and the dnsmasq module (doesn't
emit a wildcard for a named-but-uncarried peer). Not implemented.
hq cannot hold the migration's operational record — it names machines, addresses and
paths, and this repository is public — but it can say where that record is, which is what
this map is for. Asked for by the operator, who had to be told.
111, resolved: the map the control plane hands a resolution holds the machines and the
names the mesh merely serves, and the resolver's zones were given both — inventing names
under a suffix it answers authoritatively for. 112, open: adopting a tunnel gives the mesh
the peers' addresses and none of their names, so taking the resolver before they enrol
stops three machines resolving at all.
Both found by reading the plan before pushing it.
Found reviewing the resolver's conversion. Nothing fails while the node is adopted; it
fails at the flip, and it is the same split that decided which container survived the
hub's address change.
Found when adopting the tunnel moved the hub's address: the declaration followed, the
running container did not, and the forge lost its database. Issue 102's rule broken one
level down, and issue 103's fix stopping one input short.
Implemented in mesh-controller #49 and mesh-host #24. One proposal was rejected on the
record's own terms: converging the hub is not made to wait on other machines' migrations.
The addresses follow: the control plane holds the ports the node gave, a recorded build
holds no address at all, and the two forwarders that had been holding the control plane
together are removed. 097's stranded container turned out to be listening on every
interface and connected to the mesh's store — by its own old database, which is the only
reason nothing was at risk.
Both from the review of the issue-104 fix: a hand-applied file on an enrolled node is
recorded as carried and would remove the foundation, and nothing on the wire orders one
declaration against another.
Measured: AMQP is spoken in three places of the mesh's own code and in none of the
sdk or the modules; the predecessor's world is AMQP and retiring. NATS answers every
guarantee the bus relies on, durability via JetStream. Recommended: not underneath
the migration, not after it either — in parallel, one rehearsed rollout.
Two birth-address outages and a registry that would have been the third; a
container that keeps a stale environment after its file changes; a host command
that applied a converged declaration to an adopted node; the hub and the vault
without seats. And the decision the operator made under it all: the hub adopts
the predecessor's tunnel in place, key and peers and range and port.
Found checking the third cutover rather than running it. The first two were safe by
accident — both are reached through a host port, which survives a change of owner.
This is the first constraint found that decides the order of the migration.
Found at the second cutover. Carrying a value in works only for a module's own
secrets; a secret answered by the provision is minted, and six catalogue modules
take one that way.
Three modules in a row on one machine; the first was found by taking it and cost a
three-minute outage. The runbook's answer is a rule a person must remember, which is
the shape this repository says not to settle for.
The first pass answered under both ends, which review showed is the same fault seen
from the other side where two mappings share a number. Verified on the machine: the
forge is back on the port its own configuration has always advertised.
Found reading the second module's cutover rather than running it: the catalogue's
config drops a rule the machine's has, and no step puts the two side by side.
094's cause is one blind spot read from two ends, written up in its diagnosis; the fix
answers the first open question and not the other two, which become 096. 097 was found
looking at what the forge's cutover left running.
The controller kept one report per machine, replaced, so a resource nothing can
ever apply looked like a failure that had just happened, every few minutes, for
ever. It now counts identical reports and status says stuck after three.
The playbook, the README and the status skill knew five statuses; the cycle check
knew a sixth, 'fixed', and not 'wontfix'. Eleven issues sat in the sixth for weeks
with their fixes shipped, one step short of closed. They are resolved; the check
refuses the word from now on and accepts the one the playbook allows.
ADR 0069 had already placed the controller's manifest in its own repository, and the
raising design called the catalogue copy a thing to remove. The installer's step 3 was
handed that manifest by the build and step 9 read a second copy anyway.
The end-to-end design held 'the run rebuilds what it tests' for binaries and images
and not for manifests. The beds' inline copies fell into three kinds; only the first
is a stale copy. The other two are named: a mesh test wearing a catalogue module's
name (074) and a stocked runtime image the run never rebuilds (075).
A seeded file is created once (0087); the foundation filters before anything
listens (0088); a machine becomes the last thing it was told (design 05).
Each says how it is checked.
Closes issue 041 by decision and by code on the same branch: the catalogue
engine refuses a secret in a container's env, and a secret-carrying env-file
unless the container declares its reason; the controller reads all six of
its credentials from files; design 13 states the rule and how it is checked.
An audit of the six code repositories found eleven open issues fixed on main
with commits and beds to show (025, 027, 033, 036, 037, 040, 045, 047, 050,
052, 053), three partly (007, 026, 035), nine not (020, 031, 041, 046, 049,
054, 064, 065, 066) and one whose fix would live outside those repos (006).
Resolved ones name their evidence; partly ones say what remains; 041 records
that the exposure has widened since it was reported.
The installer half of the amended ADR 0085 is built and proven by the
genesis bed's root-secrets step; the foundation design closes its open item
and the installation design says what the installer does and what it still
cannot check.
A bed run from .work/<slug>/mesh-lab derives mesh-tools and mesh-sdk by
sibling path and fails at once when the directory holds only the touched
repos. Detached worktrees on main, never symlinks.
Recorded on the record, dated, before anything shipped against the sentences
that change. The vault is installed at genesis like the store and broker, one
per mesh, and holds every secret a module has for itself sealed a second time
to an operator key whose private half never enters the mesh — the break-glass
path the first version left open, without a key one place holds.
Design 24 says how; 07 and 21 say what genesis does not yet do; issue 071
names the fixed credentials the foundation is raised with today.
Design 24 flips to in-progress with mesh-catalog as its owner (playbook 04).
Starting the build surfaced two gaps the decision did not settle: a module
requiring `secret` receives exactly one value (069), and no command can
accept an operator's value into a consumer↔vault pair (070). Both opened as
issues rather than improvised around.
Also fills fixed-by on 067 and 068, which the cycle check refused as resolved
with no reference.
ADR 0084 (extends 0027) — a provision is served by a node-scoped provider the
consumer selects, defaulting to co-location; a module may instead carry a private
embedded instance that is not a provision. Design: 01-to-be/23-choosing-a-provider.
ADR 0085 (extends 0031) — a secret is a provision and the vault is the module that
provides it; a module's own local secret becomes an ordinary pair credential that
rotates through the existing machinery, while the controller's provisioning-credential
mint (0048) is unchanged. Design: 01-to-be/24-the-secrets-vault; doc 13 amended to
cross-link the non-pair secret.
Issues 067/068 marked resolved with amended-design set. ADR index regenerated;
records and index checks pass.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Store and broker seats were made ordinary modules; secret-minting is still a
privileged property of the controller that no module owns. Propose the vault
become a module that provides a secret provision (generate/hold/rotate/backup/
audit), node-scoped like every other provider (issue 067), subsuming the three
secret paths and giving local-secret rotation a home.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Naming which provider is only half of how a module gets a database. The other
half: a module may carry its own version/fork-pinned instance, module-network
only, no published port, not a provision — and the model has no word for it.
Add it as a second axis with its own open questions.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The mesh models provisions as mesh-scoped (one provider of a kind, a single
mesh-store). But node-specific services delivered to the mesh was the plan from
the start: both nodes already run their own postgres, SQL server, redis and
object store, and identity — currently single — already serves apps on a second
node. The model cannot express which provider serves a consumer, so it collapses
a deliberately per-node fleet to one. Provider scoping is a whole-mesh decision
across postgres/s3-bucket/oidc, not an SSO patch.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Surfaced while reviewing the host apply loop for the controller/host
split discussion. 065: a permanently-failing resource retries for ever
with no escalation — reported, but never raises its hand as stuck.
066: a partly-applied declaration leaves a mixed state with no rollback,
which is harmless for independent resources and unexamined for pairs
that are only correct together. Both are questions HQ must answer, not
incidents — status open, no fix proposed.
The P1-sweep PR left both at 'located'; the fixes have since merged
(mesh-controller #31, mesh-tools #10) and the bed proves them. Flip to
resolved with their fixed-by.
060: 44 of 70 modules now mesh-buildable (was 8) — the mechanical
majority, proven by direct builds and the bed. The structural
remainders: external-dependency fetch (new issue 064) and route-proxy's
cross-repo build context. 064 records the build environment's isolation
from public npm and Docker Hub.
The broker's port was accepted on input but not forwarded, so a joined
node reached a DNAT'd broker only until it restarted. Fixed in
mesh-controller (25e42b3), proven by the built-store-cross-node bed
adopting the foundation's broker (which restarts it) and the joined
node still receiving declarations.
One push leaves the mesh consistent (the 057 decision, proposed for
acceptance); the shared runtime waits for its broker (058). Fixes on
mesh-control fix/one-push-is-enough and mesh-tools
fix/the-runtime-waits-for-its-broker; the built-store-cross-node bed
enforces both.
Fixed by mesh-controller PR 30 (79a9e17), rebased onto the merged
registry-reach train and proven by a green fresh run of the no-fake
two-node bed built from that commit.
One green fresh run of the no-fake two-node bed is the proof: node2's
consumers open the store and broker the mesh built and adopted, over
the overlay. Fixed by mesh-controller PR 29, mesh-catalog PR 26, gated
by mesh-lab PR 34.
061: the broker module's provisioner never ran; its runtime container
named no command and the image default is the tool host. 062: a failed
artifact-store lookup composed the network without the registry trust,
turning a transient error into permanent silent state. Both located,
fixes on the 042/048 train branches.
Only 8 of 70 modules carry a build section; the other 62 exist through an out-of-band
build script plus placeholder rewriting, so the mesh's own build-and-deliver path has
never run for them. Found by the no-fake bed; blocks it and the migration's delivery
assumption.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Decisions were the one link the cycle checks skipped, and measuring found 19 of 70 records
orphaned — the credential flow and the module-runtime cluster among them, which is how a
stale premise about a settled decision survived in working memory. cycle.py now refuses an
accepted record nothing cites; the 19 got true homes (design frontmatter, the playbook that
implements 0021, META for the process records). The overview names the practice: spec-driven
development with provenance.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Findings of an adversarial review of the 055 fix: hub!=broker conflated, membership tested
as has-address, silent staleness, portless silent fallback, two hubs unrefused. One review
claim recorded as disputed against live wire measurements.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The flow the process overview draws — idea/symptom -> decision -> to-be design -> code ->
as-is — was enforced by nothing. cycle.py now refuses a to-be design naming no decision, an
in-progress/implemented design naming no owning code, a located/fixed issue with no owner,
a fixed/resolved issue with no fix, and a graduated research overview that does not say what
it became. AGENTS.md carries the cycle and a where-to-look table so a fresh session (or a
cleared context) finds the chain in frontmatter instead of assuming it. Grounding the check
surfaced two real gaps, fixed here: the work-ahead design named no owning code, and research
003 listed one became target twice.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The glossary's authority page still named the controller's seat the-controller in two
entries, contradicting its own seat section after ADR 0079; issue 058's heading kept the
pre-renumber 059; 055's fixed-by named branches that stop existing after merge (now merge
commits/PRs) and its located-in listed file paths where the convention wants repos; 056's
located-in named mesh-host, which received no fix, instead of mesh-catalog; and the design
layer never said the one-store/one-broker property is enforced — 07-the-foundation and the
installation table now state the seats.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Observed in the two-node bed: a cross-node runtime exits on 'timed out fetching the
broker's certificate' until the tunnel forms, then settles. The retry belongs in-process.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The store and broker were "one per mesh" by convention only. Each foundation module now
claims a mesh-scoped seat named after the server it guards — postgres/mesh-store,
lavinmq/mesh-broker — and the controller's seat is renamed the-controller -> mesh-controller
so all three follow one rule. The resolver refuses a second holder, closing 056. Glossary,
the foundation and installation docs, and the decisions index follow the new name.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
055 is proven end-to-end by the two-node bed: a consumer on a joined node reaches the
adopted broker over the overlay, its binding names the control-node, and its vhost is
minted. Three fixes — the broker's amqps port in the firewall, the broker credential
naming the overlay not the public address, and pushing the provider node after the remote
consumer arrives. The last is an operator ordering, not a code fix, and its silent-failure
edge (a cross-node consumer that never provisions and only crash-loops) is opened as 057.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
A two-node bed pins it: the firewall gap (broker's 5671 not in listens) is fixed
in mesh-catalog; the real bug is modules.go:294/build.go:231 building the broker
URL with the genesis public MESH_BROKER_ADDRESS, which a joined node cannot reach.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
"One postgres, one lavinmq" (ADR 0078) holds only by genesis assigning the
adopted modules to the control-node alone; nothing enforces it. A second assign
raises a divergent second server, silently. Open questions: a mesh-scoped
exclusive seat, or explicit adoption the controller refuses to place elsewhere.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Settles the design repository now that the self-upgrade build is on main:
- Records the two decisions that shipped without a record — ADR 0077 (the
controller/foundation/node vocabulary) and ADR 0078 (the store and broker are
ordinary modules); accepts ADR 0075 and 0076, which shipped work rests on.
- Fills issue 051's amended-design and wires ADR 0078 into 07-the-foundation.
- Sweeps the repo rename (mesh-control -> mesh-controller) into the mutable docs
now that the forge repo is renamed; updates the glossary note and repos.md.
- Fixes the six broken links from the design-doc renames, indexes the glossary,
regenerates the decisions reading order.
Both checks (records.py, index.py) are green. Statuses stay honest: the build is
on main and lab-proven but not deployed as the production mesh, so the to-be docs
remain in-progress and the as-is layer (the hal mesh) is unchanged — graduation
to implemented + as-is belongs to deployment, not merge.
https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Marks WBS 3.2/3.3/3.4 done and resolves issue 051: the foundation's store and
broker are adopted in place as the postgres and lavinmq modules, upgradeable
through their stated windows, source-tracked by status. A bare-metal mesh runs
one postgres and one lavinmq, proven 22/22 in the one-node lab. The two follow-up
gaps are tracked as issues 054 and 055.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
054: the adopted servers bind 0.0.0.0 from genesis but the packet filter is
installed later, so there is a window where they are open with only bootstrap
credentials. 055: the servers bind on the control-node and the one-node bed
cannot prove a consumer on another machine can reach them over the overlay.
Both are follow-ups to issue 051's adoption (WBS Phase 3), tracked rather than
rushed.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
"control plane" -> controller and "substrate" -> foundation throughout
03-DESIGN, 00-META and the README, with 06-the-control-plane.md and
07-the-substrate.md renamed to 06-the-controller.md and 07-the-foundation.md.
The immutable 02-DECISIONS records keep their original wording (and links to
them are unchanged) — a term retired here may still appear there, which the
glossary explains how to read.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Locks the vocabulary that kept drifting in conversation — controller (not
"control plane"), foundation (not "substrate"), node and control-node, seat /
bench / claim, package vs artifact. AGENTS.md points at it as the authority.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Records the decision the package-registry work turns on — the SDK is built on a
public base and published before the toolchain that consumes it, so nothing is
circular; mesh-tools stays the thin toolchain base but resolves the SDK by
version. Reconciles docs 12/17/22 and indexes the record.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The protocol spec claimed the envelope and grant drifted across implementations.
Inspection showed the wire agrees — envelope required headers match, the two
optional ones are legitimately optional, and the grant wire (the contributions
file) is identical on both sides. The disagreement was in dead types, now removed.
So phase 1's 'make them agree' work is done by deletion and correction. What
remains is a conformance fixture as prevention — pinning the envelope and the
contributions file so a future change that breaks agreement fails a test — and a
full per-capability suite is deferred until a third language actually needs it.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The record claimed the two implementations already disagreed. Inspection showed
the live wire agrees: the disagreeing grant types were dead (removed), and the
envelope's two extra headers are optional and set when relevant, not missing.
The danger was dead types contradicting the wire, not live disagreement — which
is a sharper reason for specifying the wire and checking against it, not a weaker
one. The model stands; the conformance suite's job is prevention rather than
repairing a present break.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The correction the operator pushed: stop running a 20-minute lab against a mesh
mid-transformation, debugging paths the next phase deletes. The base-build hang
is almost certainly the SDK resolving from a git URL inside a docker build (issue
053), which Phase 2 removes — so debugging it on the current shape is debugging
deprecated code.
Phase 0 folds in: the installer's own regressions are fixed and committed;
whether it runs green is the final acceptance test, after the phases that change
its build path are in. Faults that can be reasoned out of the code path are, by
reading rather than running.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Everything decided this cycle and not yet built. Phase 0 gets the installer
green, because nothing else is testable end to end without it. Phase 1 makes the
protocol one thing and fixes the Go/TS drift the installer's own provisioning
exercises. Phase 2 stands up the private package registry ADR 0014 assumes and
publishes the SDK into it. Phase 3 adopts the substrate so one postgres and one
lavinmq serve everything, which is the hardest and needs all three above.
Order is dependency, not preference. Each phase ends at a run rather than a
paragraph, because a phase that ends at a claim is how things went missing this
cycle without anything complaining.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The lavinmq module raises its own server container and provisions vhosts on it,
separate from the substrate's mesh-broker. A mesh with the module assigned runs
two LavinMQ servers where one belongs — the exact AMQP twin of the two postgres
containers.
The module's own provisioner already assumes one server: it creates a vhost per
consumer, named for the login, isolated by the vhost boundary — the analog of
postgres's database-per-login. So the mesh's own control traffic is the / vhost
and every consumer's broker is a vhost beside it, all on one server.
Adoption therefore means the module does not run a server of its own: its server
is the substrate broker, adopted, and the module contributes the provisioner,
tools, events and bootstrap against it. The isolation model is already built;
what remains to design is only raising the one broker at genesis and then holding
it as a module, over the broker it is.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The substrate's store and the postgres module are the same thing. They were two
rows only while the substrate was a different KIND of thing — a store raised from
a bundle cannot provide postgres-database, so anything wanting a database needed
a second server. It is visible on any mesh built today: mesh-store and postgres,
two containers, the same image.
Which name survives is settled by the naming rule. Where a consumer speaks a
protocol the interface is the protocol, and "database" is not a capability. The
control plane's own queries use distinct on and on conflict, so the coupling is
to postgres and a "store" module would advertise a swap that fails the first time
anyone tries it.
The broker collapses the same way with a different outcome: amqp IS a protocol
that several implementations speak, so amqp is a legitimate provision and lavinmq
is one provider of it.
So adopting the substrate is not only an upgrade path — it is two rows of a
mesh's module list becoming one, twice. And it leaves nothing that is a specialty
after the pivot, which is the claim the whole design rests on and is not true
today for exactly those two.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Reported as 'the host does not survive a reboot', which reads as a mesh that
cannot come back. mesh-host/packaging/ ships nox-mesh-host.service and two
companions. The installer declines to place them because a unit file is a
packaging decision, and the lab starts the host with --host-in-background, which
says in its own help that it does not survive a reboot.
So the lab run failing this was the lab being honest, and the gap is the step
that puts a shipped unit on a machine — narrower and more fixable than what I
wrote.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The first draft implied gitea and the registry could not share a machine, because
registry claims the-artifact-store at node scope and I carried that across to
gitea without asking what the claim is for.
A machine running gitea for git and packages alongside a registry serving
artifacts is an ordinary arrangement. They are different ports doing different
jobs, and nothing about one being the mesh's artifact store requires the other
not to exist.
The exclusivity that matters is mesh-wide and already expressed: provides at mesh
scope means two providers are two answers, and the resolver refuses until one is
assigned. Forbidding co-residence adds nothing and forbids something reasonable.
Whether registry should still hold that claim is left open rather than answered
from outside its manifest — it may be protecting something about its port or its
data directory that nobody wrote down.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Two questions circling turned out to be one asked twice: should gitea be the
mesh's registry, and where does the SDK come from. The framing that dissolves
both is that artifact-store is already a provision and the registry already
provides it — so this was never about replacing a component. It is a second
provider of an existing provision, which this mesh has a mechanism for and uses
for certificate authorities already.
So: two provisions, because they are two jobs. artifact-store is content
addressed, pinned by digest, no versions and no ranges — what the mesh delivers
to machines. package-registry is an ecosystem's own, addressed by name and
version — what code resolves when compiled. Conflating them is how a mesh that
pins everything ends up rebuilding one commit into two different things.
The small registry stays the provider genesis installs, not because it is better
but because of what it is: a directory and one container, installable where there
is no database and no control plane. Gitea needs both, and the pivot needs
somewhere to publish before either exists.
Gitea also provides artifact-store, so a mesh may choose it — and choosing it
answers issues 042 and 048 by adopting something that already has accounts and
TLS, rather than reimplementing them in a registry that has neither.
Moving between providers is a designed act with a verification step that is easy
to skip and is the only thing between it and a mesh that cannot restart its own
control plane.
And the bootstrap still has no package registry when the first build needs one.
Named rather than solved, so the next person does not discover it.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Every step from a bare machine to a mesh that maintains itself, in three phases,
with each step named as the installer prints it.
The point of writing it out is the shape it exposes. The installer owns twelve
steps and ends at a mesh that RUNS. Seven more turn that into a mesh that WORKS —
the shared base, a store that is a provider rather than the control plane's own
memory, the catalogue, the replay of what was built before the catalogue existed,
the control plane rebuilt through the module path, the private network with the
node actually placed on it, and the packet filter. None of those seven is the
installer's. They are things somebody types, which is why a test had to be
written to discover they were missing.
Machines arrive last, in phase three, because a machine joining a mesh that
cannot build anything proves enrolment works and nothing else.
And five things that are not yet true are named rather than implied: phase two is
manual, a second machine cannot pull what the mesh built, ADR 0014 assumes a
private package registry that genesis has not installed when the first build
needs it, the host agent does not survive a reboot, and nothing can contradict a
claim that a machine was installed this way.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
ADR 0014 is accepted and unambiguous: each module consumes its dependencies from
the private registry, the mesh's own shared library included, and a cross-package
change is publish then consume.
So the git dependency at a pinned commit is not a mechanism under consideration.
It is the shared library being consumed a way the record rules out, and the lock
naming a sibling directory is what that looks like when nobody publishes. Issue
053 is reclassified from a question about mechanism to a violation with a
direction.
What stays open is narrower and real: which software serves the private registry,
and that a fresh mesh has none when the first SDK is built — ADR 0014 assumes one
exists, and at genesis nothing has installed it.
Recorded after arguing at length for a bespoke content-addressed alternative,
against a decision that was already made and that I had not read.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
What the SDK contains, answered by exclusion as much as by inclusion. It is the
protocol and nothing else — no configuration loader, because configuration
arrives as files the mesh wrote; no API clients, because a Plex client changes
when Plex changes and that has nothing to do with any other module; no storage,
HTTP or logging, because the language has those. The test for anything proposed
is ADR 0039's: does editing it recompile unrelated modules, and does it change
often. Both, and it stays out.
Then the worked module: events in TypeScript, tools in Go, a provisioner in Rust,
a scheduled job in Python. Four artifacts, four toolchains, four processes, one
module — and each part is an ordinary project in its language depending on the
mesh SDK the ordinary way, so a laptop resolves what a build resolves.
And publishing a package as a module capability, which makes the SDK unspecial:
it is simply the first module that published a library. A Plex client belongs to
the Plex module because that is the only thing that knows when Plex changed.
Three things left open rather than papered over: which registry (the catalogue
holds verdaccio and a forge usually serves one too, and nothing says which is
ours), who may publish (a credential that does not exist), and what a range means
in a mesh where everything else is pinned by digest — a mesh that can rebuild a
commit and get a different library is a real change, and should be decided rather
than arrived at.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
A floor every implementation needs and three capabilities independent of each
other, so an SDK can implement the floor and events and be a real thing rather
than an unfinished one.
Written as a specification, which means it says what is required rather than how
anything is arranged — and says plainly where it describes behaviour that is not
yet true. Three places it does:
x-causation-id and x-schema are specified and emitted by nothing; the Go side
writes four headers and the TypeScript side declares six. A module may serve
tools and may not call them, because a caller needs a reply queue its account may
not declare. And the two implementations disagree about what a grant carries — in
TypeScript consumer is the module, in Go it is the node and the module is From.
One word, two meanings, in two halves of one mesh.
Naming those in the specification rather than leaving them for conformance to
discover, because a specification that only described what already works would
have nothing to say about the things most likely to break.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The tool runtime's manifest names the SDK as a git dependency at a pinned commit;
its lock file names a sibling directory that exists on one workstation. It builds
only because the recipe runs npm install, which tolerates a lock disagreeing with
its manifest and re-resolves from the manifest — the one command that hides this.
It matters because a lock exists to make a build reproducible and this one
describes one machine, and because it is the first thing a fresh mesh builds: the
toolchain carries the SDK and everything with code of its own compiles inside it,
so a dependency resolved differently on the build machine than on a workstation
is a difference in every module the mesh will ever build.
And it is about to be copied. Each language's toolchain will carry that
language's SDK the same way, so the shape is worth settling before there are four
of them.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Two corrections, both from the operator and both better than what was written.
An SDK is an implementation of the mesh's module protocol in one language, and
nothing more. The first draft defined it by the test it passes, which describes
how you check one rather than what one is — and leaves it sounding like a library
that helpers could accumulate in.
And the protocol is split per capability, which was missing entirely. A module
that only consumes events uses the event capability; one that serves tools uses
the tool capability; a provider uses provisioning. Nothing about consuming an
event requires knowing how a grant is answered, so an SDK need not implement all
of it to be real.
That has a precedent here: a host declares which resource kinds it can apply, and
a partial host is a real thing rather than a broken one (ADR 0005). An SDK
implementing the floor and events is exactly as legitimate, and a module written
against it is a module that does events.
Which changes what adding a language costs. A Rust SDK doing connection and
events is useful the day it exists, with tools and provisioning following when
something needs them — rather than a language being unsupported until it is
entirely supported, which is what makes adding one a project instead of a
contribution.
Conformance is therefore per capability too: a monolithic pass or fail would make
a partial implementation indistinguishable from a broken one, which is the
distinction the whole thing rests on.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
ADR 0039 settles what belongs in an SDK. It does not say what happens when there
is more than one, and there already is: the contracts are expressed as Go types
in the control plane and host and as TypeScript types in the SDK, and nobody has
felt it because both live in one repository.
They already disagree. The provision's field is "resource" in one and "Provision"
in the other; "consumer" means the module in one and the node in the other; the
envelope declares six headers on one side and emits four on both — the missing
two being x-causation-id and x-schema, the second of which is exactly what a body
needs in order to change shape without silent misreads.
That class of failure does not announce itself. Two implementations disagreeing
about an envelope do not fail to compile — they ignore each other's messages, and
a mesh where a module stops reacting looks like a mesh where nothing happened.
So the decision is to specify the wire rather than share the types, because the
shapes are the easy half. What two implementations actually disagree about is
behaviour: queue naming and durability, which headers are required and what an
unknown one means, taking identity from the sealed credential rather than the
environment, dedup on an id only the emitter can make, pinning a fingerprint
rather than trusting an authority.
And the suite is executable rather than prose, because a specification nobody can
run is a document two implementations drift from while both believe they conform.
The two existing implementations are the first made to pass it — a suite only new
SDKs must satisfy would certify every future language against a disagreement that
is already here.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
A reference table goes stale the day somebody adds a field, so this one points at
mesh-catalog/modules/showcase — a module that uses all of it — and a test that
fails when it stops doing so. Read the module when the table disagrees with it.
Two rows in the coverage survey were stale because of this week's work: systemd
units were a file plus a service, which made every author write unit syntax and
is why "process" exists; and building from source was images only, where a
bundle now names a language and lets the mesh choose the toolchain.
And the two rows at the bottom of the resource table are the interesting ones.
"action" is refused to modules outright — the link may not carry a command, so a
module needing something done ships a program that reconciles. "service"
installs no unit by design, right for software shipping one and wrong for code
the mesh built, which has none until the mesh writes it.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The first draft said the builder refused archives. It does not. An archive is
packed deterministically, hashed, published by digest, fetched by the machine and
unpacked — the whole path exists. Only the local builder used at genesis refuses
one, and deliberately: an archive is bytes that mean nothing until something
serves them, and at genesis nothing does.
What an archive cannot do is compile. Its source is a directory packed as it
stands, so shipping compiled output means compiling somewhere first, which means
a Dockerfile — the burden this document is about. The gap is not the artifact
kind. It is that no recipe both builds and packs.
Found by reading the builder rather than the manifest schema, which is where the
first draft's claim came from.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
12-a-module-repository says what a module may build and where it goes. Nothing
said how a build is MODELLED, and the model is the problem: a recipe is implicit,
singular and always a Dockerfile; a toolchain is not modelled at all, arriving as
two build arguments the module hand-writes; a language is not a concept; and an
archive is declared in the manifest and refused by the builder.
The cost is measurable rather than theoretical. Adding a module with its own code
means repeating an incantation - two ARG bases, a specific working directory so
the SDK resolves upward, the compiler invoked by absolute path because the usual
symlink is resolved away when the base is assembled, a second stage, an env var
naming the entrypoints. Most of the catalogue is unconverted, and two conversions
done in one session were each wrong twice with a working example open.
So: recipe becomes explicit with three kinds, and toolchain becomes derived from
a declared language rather than written by every author. A Dockerfile stays, and
stops being compulsory - it is right for software needing a particular base and
wrong for "compile my module's code", which is the same operation every time.
The cost is stated before it is chosen: every language is permanent, and the
contracts are already expressed twice - Go structs and TypeScript types kept in
step by hand. A second language makes that drift. So language-neutral contracts
come first, or the drift gets worse while hiding.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The packet filter generates its rules from what modules declare they listen on.
The broker is not a module, so it declares nothing, so its port is not opened.
Every machine dials that port to enrol and to receive every declaration it is
ever sent.
Invisible where it is assembled and fatal on the next machine: a mesh of one
never dials its own broker across the network, so the ruleset looks right. The
first machine to join is refused at the packet filter during enrolment, several
steps from anything that reports it — and assigning the firewall before joining
machines is both the natural order and the one that breaks.
This is 051 in a second place. That issue says the substrate cannot be updated
because the mesh holds no record of it; the same absence means the firewall
cannot know it exists. Ssh is already a floor for the same reason — a machine
nobody can reach is a machine nobody can repair — and the broker may be the same
class of fact.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The store and broker come from a bundle the installer writes once, with images
pinned in it, and nothing can change them afterwards: no build, no version to be
behind, no roll-out, and no way to report being out of date, because the mesh
holds no record of them as modules at all.
That is backwards. They are what everything else depends on, so their updates
matter most, and they are the only things with no mechanism to deliver one. A
mesh with a year-old broker reports itself entirely current.
The fix probably already exists: the control plane is carried, raised and then
adopted as an ordinary module pinned to what is running. Nothing in that pattern
is specific to the control plane. It would also remove a duplication visible on
any one-node mesh — the same postgres image running twice, because a store that
cannot be a module cannot provide a database to anything.
Recorded with the two hard parts stated rather than waved at: upgrading a store
the control plane is reading from, and upgrading a broker over the broker.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
A mesh raised from bare metal built six modules and its catalogue reported three:
exactly those built after it began running. Missing were the shared base, the
store, and the catalogue itself.
The hole is never random. On a fresh mesh the modules built before the catalogue
are by necessity the ones it needed in order to exist, so the foundation is
always what is absent, on every mesh, at the moment the graph is first populated.
It breaks the question the catalogue is for: build edges hang off the base, so a
catalogue with no record of it answers 'what must be rebuilt' confidently and
wrongly. And nothing reports the gap, because a catalogue cannot know what it was
never told — this was found by comparing its answer against what the mesh had
just been watched doing.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
A module's broker account is scoped to what it declares it emits and consumes. A
tool call needs a reply queue, which that scope does not cover and should not. So
the account is right, the request is reasonable, and no account exists that can
make it — asking a module its own question, from its own container, with its own
credential, is refused.
It matters because a module's tools are its operator-facing surface: the
catalogue serves the five questions it exists to answer and nothing can reach
them. It is also why a running module keeps being mistaken for a working one — a
test that cannot ask anything checks a container is up, and that substitution has
hidden two faults this week.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The registry says every machine pulls from it and opens its port to the mesh for
that reason. A machine that tries is refused by its own container runtime: the
registry serves plain HTTP and anything but loopback is treated as HTTPS.
It has never failed, and that is the finding. Every proof that a machine can
fetch a mesh-built artifact was a proof about the machine that built it, where
the reference was loopback. The bed that uses a routable address gets away with
it because the harness writes the runtime's configuration before the mesh exists.
Kept separate from 042, which they are easy to confuse: that one is about not
being allowed to pull, this one about not being able to whatever the credential
says. Fixing 042 alone leaves a machine with a valid account for a registry its
runtime will not talk to.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The report covered ports the firewall cannot close. The matching fault is ports
it does close and should not: rules come from what modules declare, a migrating
mesh knows about almost nothing, and anything listening on the host is dropped.
No module declares an ssh port and the generator has no allowance for one. The
session that loads the rules survives on conntrack until it drops, and then the
machine is reached from a rescue console.
Rehearsed on three lab machines before doing it on an anchor: loading the mesh's
rules refuses an ordinary host port and leaves a published container port
reachable, same prober, same second.
It is deliberate — no forward chain, because dropping there would stop every
container — but traffic to a published port never reaches the input chain, so
the firewall is silent about it. The anchor publishes 38 such ports today, all
filtered by the system being replaced, so the cutover would open every one of
them while reporting a firewall that is up.
Found while migrating the first module. Five approaches were tried and observed
to fail, including resolving the index and naming the platform at both ends; a
speculative fix was written and reverted rather than shipped, because it did not
make the mirror work.
One module uses this today, which is why it went unnoticed — and it is the shape
most of a migration wants, because the services being moved are third-party.
The design said how the builder arrives was unsettled and that nothing
installed it — the one gap stopping a fresh mesh from producing anything. Both
are now false. What is still true is narrower and worth keeping separate:
nothing asks a raised mesh for the rest of the catalogue, and no bed asserts
that it could.
The mesh could already say which modules a base change invalidated, and could
not do anything about it: each recipe named one particular copy of the base by
fingerprint, and rebuilding produced the old one. Worse, the copy each named
existed only inside a throwaway lab, so those three modules could not be built
anywhere at all — and the line each replaced had the same fault.
The builder's arrival was the one rule the document said nothing checked. It is
checked now, by both genesis beds — and so is the thing that distinguishes a
built control plane from a carried one, which every earlier assertion accepted
either way.
The section saying the change was decided and had not happened now contradicted
the section below it. It also records the argument that failed, because a reader
will otherwise ask the same question and reach the same wrong answer.
Written an hour ago claiming a produced image must be published before anything
can fetch it, so the registry had to precede the control plane. The premise is
false: the machine that builds the image is the machine that runs it, and the
temporary control plane names a built image exactly as it names a carried one —
by the digest of its own configuration, which requires nothing to have served
it. Building changes where the bytes came from, not where they are.
Rewritten rather than superseded because nothing has been built on it and
nobody has read it: a record that contradicts itself is a draft, not a decision.
The argument is kept, because it was asked for and a negative answer is the
result.
The two questions the design record named as the one gap stopping a fresh mesh
from producing anything. They cannot be answered apart: a builder with nowhere
to publish has made a file on a disk.
The registry's role did not change — the answer to 'must it precede the control
plane' did, because the control plane's image is now produced rather than
carried, and a produced image must be put somewhere before it can be fetched.
Cloned alone, built by the mesh, every module moved onto it, and the base then
changed for real: all three went stale naming what moved. A comment-only change
correctly makes nothing stale, which found a second bug — staleness compared
commits where it should compare artifacts.
The runtime every module compiles against cannot be built by the mesh, so the
one rule that would catch it moving can never fire. And a container keeps the
values it was created with, so two good applies can leave it running on neither.
Corrects ADR 0070, written an hour earlier, which had the control plane consuming
the catalogue in order to compose a declaration. That was written before the two
graphs had been told apart and creates a dependency that need not exist: a
catalogue that is down would leave the control plane unable to compose the thing
that would repair it.
The catalogue links module-versions to each other and does not know nodes exist.
The control plane links module-versions to nodes, and holds capabilities and
claims. They meet only when something is installed, and everything between them
travels as events over the broker, on durable queues, so nothing is lost when a
receiver is away.
Build order is not computed anywhere. The builder never consults the graph and
builds what it is asked for; the catalogue asks for the next build after the
previous registration, so ordering holds by construction. Its rule is a condition
rather than a schedule — rebuild once everything a module was built against is
current — which covers a chain and a diamond alike.
Left open: whether an upgrade is applied or merely noticed, and whether a module
on several machines upgrades on all at once.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
ADR 0070 has the init builder clone the source and does not say from where, and
ADR 0067 had rejected building at genesis partly because the forge runs on the
mesh being rebuilt. That objection binds only when those are the same mesh, which
is true exactly once.
So genesis clones from a mesh by name, and if that mesh is lost the name moves to
another that holds a copy — recovery is a name pointing elsewhere rather than a
backup being restored, and every installation adds somewhere it could point.
It names a commit and checks what it got, because the forge a mesh installs from
is the trust anchor for everything that mesh will run. On 2026-09-11 that forge
was running a cryptominer and tampering with git operations in flight; nothing was
altered, but a mesh installing during those hours could not have known that.
What relationship a mesh keeps afterwards is left open and named, so that whoever
writes the init builder does not settle it by accident: a snapshot and then
independence, or a continuing upstream for core modules.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The graph had no owner: what modules are, what they require, what they claim and
what they are built against all sat in the control plane because that is where it
was written first. The control plane's own test says otherwise — anything a single
machine could answer alone is not its work, and what a module needs requires no
knowledge of any node.
So the catalogue becomes a core module beside the control plane and the builder,
owning the graph and serving tools over it. The control plane consumes it, which
is the opposite of what the tiers suggest and is therefore stated rather than
inferred.
That makes the catalogue a fourth thing that cannot arrive through the ordinary
path, so the claim written this morning that the list was closed at three is
corrected. All four are answered by one mechanism instead: the installer carries
an init builder and the core modules are built on the machine, so what is carried
is a builder rather than a result and nothing is left without a route.
Two things are open and named rather than assumed: where the init builder clones
from, given the forge normally runs on the mesh it would be rebuilding, and what
it publishes into, given the registry is installed later in the order today.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
A build machine was refused the build queue, and this was raised as a gap in what
a manifest can express. It is not: `builder issue` creates exactly that account,
three lines from the code being read at the time.
Kept rather than deleted, for the one real thing in it — the wrong verb succeeds
and reports success, producing an account that authenticates and can do nothing,
so the failure surfaces a layer away as a permissions error that reads like a
missing feature.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Raising a scenario occupies the machine and the person who started it, and
running in the background against a working copy is worse than waiting: a run
reads that copy as it goes, so editing while it runs yields a result describing a
state that never existed.
Proposed rather than accepted. The load-bearing part is the restriction — a
request names a bed and a commit and nothing else — because a request that could
say what to install and where would make the lab a second way of installing a
mesh, which is the arrangement that just cost a year of late-found faults.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The design said a module's manifest sits at a repository's root, full stop, which
means one repository per module. Nothing that exists is shaped that way: the
catalogue holds sixty-seven modules one to a directory, no code repository has a
manifest at its root, and the system being replaced has always built a module
from a repository and a path.
So the builder could be asked to build nothing that exists — pointed at the
catalogue it finds no manifest, pointed at a module's source it finds none
either. Recorded as a decision because it moves the core modules' manifests
beside their source, and corrects the design that said otherwise.
Also corrects, in the same document, how the three things the build loop cannot
produce actually arrive. They were written as though all three were carried in.
Only the control plane is: the registry is pulled from the public internet, and
the builder has no route at all — which is now stated as the open one rather than
implied to be solved.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
The one complete account of standing a mesh up was an integration test, and a
fixture is free to invent what it needs — which is how a registry that exists in
no production hid two faults for as long as the lab existed.
Written from what the installer does, not what it should do: genesis and joining
are separate moments, the lab runs the installer rather than describing
installing, and three things that are not true yet are named rather than glossed,
including one rule nothing checks.
Claude-Session: https://claude.ai/code/session_01D6qtiYU3P9jk3pnAXyAFyx
Found by deleting the lab's registry and giving the machines a real path out:
public images fetched, the operator's own could not be fetched at all. There is
no provision for a registry credential, no manifest field, and no step in
enrolment that establishes one.
It applies to the mesh's own store too, which today asks for nothing — a
decision that has never been written down as one, and so reads as an absence.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Both carried 'model access', which is not one of the six the reading order
knows, so neither had a place in it. 0050 — the record they extend, on the same
subject — is 'what runs on it'.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Nine references now name the digest their tag resolved to. What closed is the
immediate fault; the open questions stand, because a digest in a repository is
wrong the moment anybody rebuilds — which is the reason the design wants the
repository to name artifacts and the mesh to hold digests.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
0066 was proven on a four-node bed before it was ratified: one node setting
moved an entire domain, a routed name resolved inside the mesh and was issued a
certificate by the internal authority, and TLS verified against that authority
with no override. 08-connectivity rests on it and could not while it was
proposed.
0067 records what deleting the lab's registry exposed — that pinning quietly
required a registry before the thing that lets a mesh have a registry could
start — and the pivot that resolves it.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Both carried a topic outside the six the index knows, so neither had a place to
be read in — 0067 had none at all. Both are 'the tiers', beside 0036 (bootstrap
ends at a usable mesh) and 0007 (connectivity), which is what they extend.
0067 cited 0041 for tier 0's property; on this trunk 0041 is events, and the
record it meant is 0005. A citation that resolves to the wrong record reads as
corroboration, which is worse than a dead link.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The doc's own rule is that the lab verifies capability by outcome, never by
reading a setting. A path out is exactly that kind of claim — a route and a
policy can both read correctly while nothing gets through — so it is asserted by
fetching something, in the table with the rest.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A container runtime on the same workstation sets the kernel's forwarding policy
to drop, and the virtualisation daemon's own accept rules do not override it —
both are consulted and a drop anywhere is the answer. The machines then get
addresses and resolve names, because the daemon's resolver is on the bridge, and
discard every packet to anything real.
The sharpest 'available is not adequate' yet: nothing is misconfigured, nothing
logs, and it presents as every image pull hanging. A workstation that runs
containers is the ordinary case, so it is a prerequisite — verified by reaching
something, never by reading a setting.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
029 says a module providing the artifact store may not build artifacts. The
pivot shows that is the narrow case: it may not require anything the store is
needed to deliver. A route-label migration gave the registry a public name and
a route requirement, and at genesis nothing provides a route — nor can anything,
since the routing stack needs images and images need the store.
The same cycle through a door the existing wording did not cover, so the rule is
widened where the bootstrap decision states it, with a check that would catch the
next one where it is written.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
There is no installer. The complete account of standing up a mesh is an
integration test in the lab, and a fixture may invent what it needs — this one
raised a registry no production has and rewrote every image reference through
it, concealing both 039 and the fact that a first node outside the lab had no
bootstrap path at all. The bed being green said nothing about whether a mesh
could be installed.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Nine images across seven modules name a tag, not a digest. ADR 0006 forbids it
and the host refuses it by name — and the refusal has never fired in a bed,
because the lab pushed every image into its own registry and rewrote every
reference to the digest it had just assigned. The harness was supplying the
property under test. Found by deleting the harness.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The control plane's image is built from source and pushed nowhere, so it has no
manifest digest; a registry assigns those. Pinning therefore required a registry
before the thing that lets a mesh have a registry could start — a dependency the
rule created by accident. The lab hid it by raising a disposable registry no
production has.
Proposed, not accepted.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
0056 is 'the authority is the control plane, not a database', drafted on the
in-progress record chain this branch was cut from before those three records
landed. Two files would have collided at merge, which is the kind of thing that
is cheap now and confusing later. The code written against it still says 0056
and is corrected separately.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The real defect is same-node consumers announced the declared port instead of
the assigned/published one; the loopback observation was a stale pre-0038 build.
Fixed in mesh-control fix/same-node-provider-announced-port (c147a26) with a
regression test.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A from:mesh provider is announced (per 018's fix) at the node's private-network
name, but its port is published bound to loopback, so consumers dialing the
announced <node>.internal:port reach nothing. Diagnosed from the lab: DB
consumers that require the database at startup crash-loop; the provider is
healthy on loopback only.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A route contribution carries a label; the node carries its public domain; the
mesh composes <label>.<public-domain> and holds no name map. A granted route is
published into internal resolution so anything in-mesh (notably an internal ACME
authority) can resolve and reach it. That authority certifies routed names by the
same path a public one would, differing only in issuer and trusted root.
Records the decision as proposed and amends connectivity SS2/SS3/SS5 plus its Open
list with the lab findings behind it.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Extends ADR 0050: model-access, a provision, may be answered by a NODE that
hosts a model (Ollama/vLLM serving an OpenAI-compatible endpoint) as well as by
a licence record. A node-answer delivers an endpoint (base URL + model, and a
key only if the server wants one), rides the ordinary serves/binds path, and
uses no adapter; the resolver already prefers a local model over a licence. One
provision, two answers — the mesh's own model sits behind the same interface as
a vendor's. Names the mesh-scope-alongside-licence limitation and its fix.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Closes the usage-tracking half ADR 0050 left open. One row shape (0050's), recorded at two
consumer grains: the holding module (licence-level, e.g. Anthropic utilization%) and the agent
session (per-session tokens/cost, since a session IS a consumer per ADR 0026). Produced by the
vendor adapter; read on a schedule (0053); recorded as events the audit-logger keeps (0041/0042)
plus a queryable usage context store (0008); usage is not a credential and is recorded in the
clear. Reuses the session, schedule, event, and store the mesh already has. Accepted per the
user's choice to build the full feature incl. per-session token/cost.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The recurring twin of run-once (0052): a schedule modifier on the container shape, reusing
its security bound (no new host action/shape, strictly less than an action) and reversing its
gating rule — a scheduled step runs after convergence, does not gate the apply, and a failed
run is logged, not fatal. Unblocks kometa's sync and pollers. Accepted per direction to build
the primitive now.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The run-once lifecycle primitive: run-once:true on the existing container shape,
gated by declaration order + exit 0, idempotent by digest. No new host shape, no
arbitrary command — strictly less powerful than an action. Resolves issue 037.
Verified sound and faithful (ADR 0005/0047); status proposed->accepted.
A module can declare state but not a step that runs. mosquitto must seed its
dynsec admin into dynamic-security.json before the broker starts, or the plugin
aborts; the database providers need the same for first-boot migrations and
health gates (04-ISSUES/037). The old event-hook engine that did this was
powerful and flaky; this is the narrowest sound mechanism instead.
A run-once step is an ordinary container marked `run-once: true`: the host runs
it to completion, requires exit 0, and gates the apply on it — so what the
declaration places after it (the broker) starts only once it has finished.
Gating is by declaration order, not a resolved dependency (ADR 0005); the
completion marker is the recorded declaration digest (ADR 0018), so a re-apply
does not re-run it unless the declaration changed. No new host shape and no
arbitrary host command: strictly less powerful than an `action`.
Points 04-ISSUES/037 fixed-by/amended-design at the record; index regenerated;
records.py and index.py pass.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Your decision, ratified: the media library (and shared/pre-existing data) is
operator-owned and external; a module declares access, not ownership; the host
mounts but owns nothing (no create/chown/reconcile/remove); an absent accessed
path is refused clearly; several accessors co-resolve. Status proposed->accepted;
index regenerated (records + index checks pass). Implementation lives on the code
branches (mesh-control/catalog/host), held for merge after the convergence fix.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Resolves 04-ISSUES/036: eight media modules each declared the shared
library and download directories as their own resources, and the
resolver's duplicate-owner refusal — right in general — would refuse the
stack's only sensible assignment the first time two landed on one node.
The decision, from the operator: shared, pre-existing data is
operator-owned and external. The mesh does not create, chown, reconcile
or remove it. A module declares it needs access to such a path (read or
read-write); the host mounts it and owns nothing. Several modules
accessing one path is normal — the duplicate-path refusal is about
ownership, not use. An accessed path absent at apply is refused clearly,
not created. Extends ADR 0030: the third case the host had no word for,
what it neither made nor configured and must not touch.
Point 036's fixed-by/amended-design at the record; mark it located in
mesh-control, mesh-catalog and mesh-host. Regenerate the decision index.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Turn the completed vendor-agnostic analysis into HQ design. The model-access
provision stays one vendor-blind interface (extends 0024/0027); the
vendor-specific lifecycle moves into a per-vendor adapter keyed by the licence's
`vendor` field, mirroring registrar-scoped public-dns providers (0044), named at
the consumer's real coupling per 0040.
The crux is the sealing-vs-central-rotation carve-out: for refreshable-grant
vendors only, the manager node holds the refresh token encrypted at rest (a
bounded, declared exception), access tokens sealed per holder, refresh stripped
on delivery. Static-key vendors keep full sealing.
Amend 03-DESIGN/01-to-be/14-model-access.md with the adapter generalisation as a
proposed section (prose + diagram, no code); regenerate the decision index.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Init's 003 was already resolved with its own consolidated attribution
(03-DESIGN/01-to-be/08-connectivity.md); the re-homing overwrote it with the
session's ADR-0045 firewall attribution. Init is canonical and the firewall
decision is recorded in the ported ADR 0045 regardless, so 003 is restored
untouched. No initialization record is modified by this reconciliation.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The real work lived on initialization (consolidated decisions 0001-0038, the
fuller issue set 001-031, the control-plane/substrate/node-lifecycle/delivery
design, research 011/012, the checks tooling). main had diverged onto a stale
base and only carried this session's genuinely-new work. This merge makes
initialization's tree canonical on main; this session's 11 new ADRs and 6 new
issues are re-homed on top in the following commits. initialization is recorded
as a parent so its history is preserved in main's ancestry.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Surfaced converting the catalogue: a module can declare things that exist
(dir/file/network/container) but not a step that runs at a point in its
lifecycle. mosquitto's dynsec admin client must be seeded before first start;
the DB providers have nowhere for a migration or health-gate; it is the timing
face of issue 011. Framed as a missing module capability, not a defect. Records
the prior-art event hooks and their real warning — powerful but complex and
flaky — so the resolution avoids rebuilding that. Ends in open questions
(run-once resource vs general lifecycle hook, where the code runs, idempotency).
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Numbers 008 and 009 were taken on main after this branch was opened
(008-provider-runtime-has-no-seal-key, 009-runtime-config-change-does-not-restart,
both merged). Renumber the seed-file-wipe and shared-directory issues to the next
free numbers so merging records two more issues rather than duplicating two.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Records the branching-and-merging workflow for code changes across the mesh
repos, written against a failure it names: branches and MRs opened per unit of
thought, treated as done when opened not merged, and named differently per repo,
so they pile up unmerged — one session left sixteen to consolidate by hand. The
rule is one feat/<slug> shared across every repo a feature touches, isolated in
.work/<slug>/<repo> worktrees off main, pushed and opened as one MR per repo only
when the whole feature is done, then merged promptly. Adds the ground-rule
pointer in AGENTS.md and the row in the process overview.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Brings the independent ADR branches (0044-0052) onto one branch so hq lands as a
single MR, and ratifies the five that were still proposed — 0017, and 0049-0052,
which are implemented and green in the lab. With 0053/0054 already accepted here, the
whole ADR chain 0044-0054 is accepted on this branch.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
ADR 0054 accepted with option E (a declared slug). Issue 010 resolved: the login fits
via the slug, and the minted secret shrinks to 40 chars for S3's secret-key limit —
both halves of an S3 credential now fit the tightest backend.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Implementing A (bound the identity at 20) revealed the readable budget is node+module
<= 14 chars — so tight that the catalogue's own test names (workstation+keycloak, 25)
compact to an opaque hash. B's fallback would fire for the common case, not the rare
overflow, inverting A+B into mostly-opaque identities. Option E — an optional short
slug a module/node declares, preferred over the cleaned name — is the escape hatch B
wanted to be without the opacity: legible because a person chose it, and it makes an
early refusal palatable (refuse on the slug field, not the machine's name). B dropped;
E recommended over a bound of 20, composing with C later if needed.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Sketches the options for issue 010: the mesh's identityLimit (63, postgres's) is not
the shortest among the backends the derived name reaches — S3's is 20 — so CheckIdentity
lets an over-long access key through and minio fails at provision time. Options: bound
by the true minimum and refuse at assignment (recommended, with a compact fallback held
in reserve), per-interface bounds, or a provider-generated identity (rejected — breaks
"the mesh says the identity once"). Links issue 010 to it.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Found doing the per-backend provider e2e: redis and postgres accept the mesh's `as`
(mesh_<node>_<module>) verbatim, but minio's S3 access key is capped at 20 chars and
`as` is 22, so the provisioner cannot create the service account. `as` is doing two
jobs — a stable identity the two ends agree on, and a literal identifier a backend
must accept — and those are not always the same string.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
ADR 0053 accepted; adds the scope boundary the umami rework surfaced (credential
provisions vs data provisions — analytics' generated siteId return is left to a
separate decision) and records the lab proof. Issue 008 marked resolved: the sdk
harness and the four adapters are reworked, the symmetric seal removed, and
provider-uses-mesh-credential is green.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Issue 008's trace confirmed the premise in control-plane code: the mesh already
mints one password per consumer/provider pair and delivers the provider its copy
(SecretFor/SecretsFrom/grantsFor -> Grant.Sealed; the receives contribution carries
As + Secret). The provisioner's symmetric seal is an orphaned, contradictory second
model. ADR 0053 corrects the provider contract in one place (the sdk harness):
providers create the resource with the mesh-supplied login and password and drop
seal/key/return entirely. Reframe 008 as contract-first (every provider, not four).
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A cross-repo trace showed nothing writes the provisioner's grant-request files,
nothing reads its sealed credentials, and no consumer unseals — while the mesh
already mints and delivers provider/consumer credentials asymmetrically with no
shared key. The fix is to drop the symmetric seal and have providers consume the
mesh-minted password, a breaking provider-contract change that wants an ADR.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Found rolling the runtime out to the catalogue: a module's runtime reads its
settings-merged config file once at start, but a container is only recreated on a
spec change, and file content is not part of the spec. So updating settings
re-renders the file and nothing re-reads it — ADR 0051's "on the fly" holds only
for config set before first start. Services have restart-on; containers do not.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Found building the module-runtime vertical slice: a provider's provisioner
requires a seal key it has no way to receive, and the consumer no way to obtain
the matching one. The runtime cannot come up as delivered.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The runtime-model gap the review found. A module with tools or events runs one
container — the tool runtime carrying its code — holding the one scoped account
ADR 0048 gave it. A node-wide runtime can't: it would hold the union of every
module's permissions, the isolation 0048 draws. So per-module: one module, one
process, one account. Tools served per key (serve.<tool>) so a caller names a
tool and only its module answers (superseding a shared tools.invoke); events in
the same process under the same account; the runtime image is the tool runtime
plus the module's code (the audit-logger's shape, made the rule). A plain
service module runs no such process. A provider's provisioner is a runtime too —
which is why a provisioner that emits must carry a broker credential or not emit.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A module is assigned to a node (there is no mesh assignment; 'mesh' is a scope).
The manifest is what the module IS, plus defaults; the configurable values are
settings, carried by the assignment — per-node or mesh-wide, applied at
resolution, changeable live (what a meshboard edits). Extends settings from a
config file's content to the manifest fields marked settable: foremost
listens.from (postgres from:mesh by default, from:anywhere per node — the
firewall follows), and a provider's own config (a registrar's zone/domain/
ingress). Static config in a manifest is config in the wrong place: it cannot
vary per node and cannot change without a rebuild.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The manifest refuses unknown keys (DisallowUnknownFields), 'from' is the field
that scopes a port and it is rendered to nftables (AsNftables), and the firewall
module applies the rule set. The chain from a declared scope to a dropped packet
is closed. Amended-design: ADR 0050.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
0049: a public name is provisioned like any capability — a module requires
public-dns and contributes its host; a neutral interface answered by
registrar-scoped providers (cloudflare-dns, route53-dns) that create/remove
the record pointing the name at the mesh's public ingress. Pairs with route
(the proxy) and a public cert (the proxy's ACME).
0050: answers the firewall question. The firewall is NOT a provider like the
proxy — it is a machine's own filter, derived by the host as the sum of what
its modules declare they listen on, with 'from' the whole of public-vs-internal.
Enforced both ways, unknown keys refused — closing 04-ISSUES/003. A public
service is exposed through the proxy (listens from:mesh + requires route), not
by opening its own port.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Events (0046) and their wire (0047) left open how a module reaches the
broker. The code has no generic module broker-account: only node and
builder scopes exist, so emits/consumes are enforced by nothing — a
manifest declaring a scope the broker does not draw (04-ISSUES/003).
Decides: on assign, a module gets a broker account whose permissions ARE
the manifest — read on mesh.events + its own queue bound to consumes;
write to mesh.events under module.<self>.* only; nothing else. Consuming
'#' is a deliberate, auditable grant. The account is what makes the
declaration a rule the broker enforces, not a comment.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
The wire contract ADR 0046 left open: two topic exchanges (mesh.events,
mesh.rpc, kept apart so # is a clean audit); the routing key as the event
type namespaced by origin (module.*, mesh.*, node.*); metadata in AMQP
headers (required x-event-id/x-source/x-node/x-time/content-type; optional
x-causation-id/x-schema; unknown x- headers ignored) with the body only the
payload; persistent messages; per-consumer durable dead-lettered queues
with prefetch; at-least-once with idempotent consumers (no false exactly-
once). The precedent is ADR 0043 for declarations.
Supersedes the sdk's first cut (metadata in body -> headers); that and the
queue config are code to align in mesh-sdk and mesh-tools.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A module emits and consumes events, both declared (emits/consumes),
parallel to provides/requires. Events are 1:many, broadcast, credential-
free — no provisioner, just the broker's topic routing — so most inter-
module reaction should be an event, not a provision. Every event carries
source/node/time so it is auditable; the audit logger is just a module
consuming '#', no privilege. A consumes for an event nothing emits is a
dangling edge and refused, like requires. One per-node runtime serves
tools, provisioning and events alike.
Extends ADR 0045; builds on ADR 0001 (the broker) and 0044 (emit/on are
stable sdk surface; the binding and runtime are not).
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
A module is one self-contained piece of software the mesh installs and
manages; the software is its identity, and capabilities/seats/provisions
are the relationships between modules, not what a module is. Records the
three relationships (shared seat, exclusive seat, provide/require), that
interfaces are mesh-owned and providers adapt to them, and the naming
rule: draw the interface at the consumer's real coupling — neutral where
the coupling is thin (analytics), protocol-scoped where the consumer
speaks a protocol (postgres/mssql/mongodb), never false genericity.
Supersedes 0017 (domain grouping — wrong axis), refines 0002, generalises
0027's protocol-not-product rule.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Supersedes ADR 0030's 'types, not behaviour' line for mesh-sdk. The
boundary is change-frequency, not kind: the SDK holds the stable spine
(tool-serving harness, messaging/event framework, contracts, core
primitives) and refuses per-module clients, per-module tool code, and
anything volatile — because those are what turned hal/sdk into constant
maintenance and made every edit rebuild every module.
States the rule (frequent AND cascading is the disease), why the root
cause was intra-module feature-sharing leaking into inter-module
coupling, and where per-module shared code lives instead (in the
module — a shared file, or a module-local sdk for the few large ones).
Updates repos.md's canonical mesh-sdk description to match; leaves 0030
untouched (immutable).
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Both from the 2026-09-02 review of the catalogue examples, and both
design gaps rather than defects in a file: a seed file the host
reconciles back to empty over the grants that grew in it, and a module
stack refused co-assignment because six manifests each own the
directories they exist to share. Fixing either in place would have
been picking an answer the records do not yet hold.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Declarations queue and the host applies all of them, oldest first, so a
machine pushed five things in a minute spends five applies becoming the
last one. Correct every step — each declaration is the whole machine —
and wasted in all but the final step.
Only visible since a report names its declaration: the reports arriving
were about ever-older ones, while timestamp comparisons used to happen
to pass. The fix is consumption order, not the queue; the open question
is what a superseded declaration's report should say, because silence
reads as disobedience and "applied" would be a lie.
The five declarations cover relations; a module is more than its
relations. Add the facet-by-facet coverage table so the effort cannot
conclude while tools, verification and contributions are unplaced —
and weigh each candidate gap rather than adopting it: contributions
probably dissolve into declared resources, mandatory verifiers risk
trivial ones, and the tool surface is the one facet with no home.
Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
Composing a declaration signed the machine's certificate anew each
time, and a signing carries a fresh random serial — so what the mesh
would send differed from what it had sent by one byte, for ever, and
every machine carrying a certificate stood eternally waiting.
Third find of the same rule: issued once and kept. The port had it, the
secret had it, the certificate composed fresh on every asking — and the
keeping column had existed since the serving-key migration, written by
nothing, the same shape ReleasePorts was found in.
Found by keeping the scenario standing and diffing two plans seconds
apart: one line, where four theories had none.
The registry module names its image by digest, the way the bundle names
the three a first node starts from, and goes in as a manifest. The lab
now walks that path and passes.
A manifest that provides the store and also builds artifacts is refused
at parse. The lab had been passing only because its Docker Hub stand-in
quietly received the push — a prop covering for the thing under test.
Installing the module that provides `artifact-store` requires something
that provides `artifact-store`: its image is mirrored in, mirroring
publishes to the store, and the builder refuses to run without one.
Never seen, because the lab always has a registry standing before the
mesh asks for one, and so does any mesh built on a machine that already
had one.
What it blocks is larger than a registry. The store holds images and
packed archives both — it is the module catalogue in artefact form — so
until it exists a mesh can run only what its bundle already carries.
The substrate record already answers it. It asks of each candidate
whether it can grant itself the thing it provides: the store cannot
create its own database, the broker cannot create its own virtual host,
and the registry cannot grant itself a repository. So the registry
module names its image and is never built.
With a limit worth saying out loud rather than discovering: a module
providing the store may not build artifacts of its own, a UI or a tool
server included, because there is nowhere to put them until it runs.
Such a registry is two modules.
Enforcing "declare what you mount" refuses the builder, which mounts the
container runtime's socket. That socket is not the builder's data: it
exists already, the machine owns it, and declaring it as one of the
module's directories would be a lie the host would act on.
Two kinds of mount are spelled identically today — the directory my data
lives in, and a machine facility I was granted. Until a manifest can say
which, the fourteen declared mounts are right by coincidence, which is
what this issue was opened about.
Recorded rather than decided: separating them is new vocabulary, and
inventing it to turn a check green is how a mechanism nobody chose ends
up load-bearing.
The fourteen undeclared mounts are declared. More to the point, a
manifest that does not declare one is now refused: they were right by
coincidence, and a checklist nothing enforces is a checklist that is
true until the next commit.
Refused in the control plane, because the machine cannot tell the
difference — asked to mount a path that does not exist, it makes the
directory, which is a thing it is perfectly able to do.
A module now says its port once, in `listens`, and the container's
mapping, the rule set and what a consumer is told are all derived from
one assignment. The three hand-written copies that agreed only because
one person wrote them are gone.
The half that made this an issue rather than an inconvenience was that
the substrate is not a module: nothing in the mesh had heard of its own
store, so it handed a database module the port the store already had.
The machine now says what it carries, and the mesh assigns around it.
Ports the protocol fixes became claims, which needed no new mechanism —
the mesh already had one for what is singular on a machine.
Left open, and unchanged by any of this: whether a module should publish
to the machine at all. Assignment makes publishing safe without making
it necessary.
Jochen's call, and the right one: a module cannot choose a port well,
because it is written once and assigned anywhere. Any number it picks is
a guess about a machine it has never seen, and two modules guessing the
same number is not a mistake either of them made.
Writing it up turned up something the issue had missed. The same number
appears three times in every module — the rule set, what a consumer is
told, and what the runtime publishes — and nothing checks that they
agree. They agree today because one person wrote all three. A module
whose `serves` said one thing and whose container published another
would resolve, compose, apply, and hand every consumer a port that
answers nothing.
So the decision is one source with the other two derived, and an
assignment made once and kept, as a credential is.
The part that needed thought is ports that cannot move — mail on 25,
submission on 587. Those become claims, which is what the mesh already
has for what is singular on a machine. Two modules wanting 25 is the
same shape as two wanting the seat, and gets refused by name at
assignment rather than by a container runtime at apply. That makes this
mostly a matter of pointing an existing mechanism at ports.
Left open: whether a module should publish to the machine at all.
Assignment makes publishing safe without making it necessary.
Found by fixing 027 and pushing again. The declaration is now accepted
and the database container still cannot start: the mesh's own store
holds 5432 on that machine, and the module publishes 5432.
Nothing catches it because the substrate is not a module. It arrives
from the bundle before there is a mesh to ask, so the control plane has
never heard of the store and does not know it holds a port. Resolution
can compare modules with each other and cannot compare one against what
the mesh is built on.
Nor does it compare modules with each other. A port is exclusive on a
machine in exactly the way a claim is, and the mesh has a mechanism for
that which ports do not use.
It has been met before: the end-to-end test that exercises a real
database publishes 5433 rather than 5432, inline, with nothing saying
why. That is how a constraint becomes folklore.
The open question is bigger than the bug. Whether a module should
publish to the machine at all decides how a consumer reaches it, and
changes what `serves` means.
Found by the forge failing to start. I had put `restart-on` on nine
containers so they would pick up a rotated credential; it belongs to a
service, and the host refused the whole declaration.
Removing it fixes the modules and leaves the reason I reached for it.
The mechanism is written against exactly this, in the host's own words:
a running service does not re-read its configuration, so replace the
file, find it already running, do nothing, and the machine keeps
behaving as before while every check passes. Every word of that applies
to a container, and nearly everything the mesh runs is one.
The cost is concrete. Rotation replaces the file and tells the provider
to accept the new credential. A provider reconciles, so it takes it. A
consumer is usually a container, so it does not — and the two ends hold
different passwords, which is the fault ADR 0001 records costing two
days. The test that proves rotation works uses a consumer that reads the
file on each attempt, so it does not meet this.
Two things a fix has to keep: it stays declared state rather than a
command, because the link may not carry an action; and where an env-file
changed, the honest verb is recreate rather than restart, because a
container's environment is fixed at creation.
Three corrections, two of them to things I wrote today.
026 is the serious one. Four modules mounted fourteen host paths nothing
declared — the mail spool, the databases, the object store's data. The
runtime creates those as root, so owner and mode go unapplied, and the
rule that keeps a directory holding data the mesh did not put there is
written in terms of declared directories. It reached the configuration
and missed the data. The cause was carrying compose files across: a
container shape that can express one gets filled in like one.
025 claimed nothing turns a tag into a digest. That is false, and the
answer was designed and built before I wrote it. A module names an
artifact, not an image, and `kind: upstream` mirrors somebody else's
image into the mesh's own registry, pinned by the digest it lands with.
The two-document split the issue described as the shape of a fix is the
design. Pinning twelve images by hand was treating the symptom, and left
them pointing at a public registry rather than the mesh's.
And the image store was written up as something the mesh does. It is an
ordinary module — considered for the substrate and removed, because the
test is whether the control plane needs it before its first instruction,
not whether it can grant itself one. So somebody's own registry is the
same module as the mesh's.
Recorded against phase 3, because the phase note said running them
needed images stocked and provisioners built — and missed that not one
of them named an image that exists. Sixty-four zeros where a digest
belongs, eighteen times, parsing and resolving perfectly.
The forge now runs: on a database another module provides, with a
password it did not choose and a connection string it could not have
written itself. First of these descriptions to be started rather than
planned, and it exercises the whole of the credential work.
What remains is the mechanism rather than the data. Nothing turns a tag
into a digest as part of the mesh's own work, so it was done by hand —
which is what the issue says a person should not be asked to do. Asking
a registry takes a second and pulls nothing, so the main argument for
leaving it undone is gone.
The refusal landed, and the examples pin images that exist. What is
still missing is the part that makes it unnecessary: nothing in the mesh
turns a tag into a digest, so it was done by hand — which is precisely
what the issue says a person should not be asked to do.
Recorded because the resolution mechanism turned out to be trivial:
asking a registry what a tag points at takes about a second and pulls
nothing. That removes the main argument for leaving this open.
Also records the two faults that fell out of pinning for real. The mail
system named seven repositories that do not exist, because it publishes
to a different registry than the manifest assumed, and one of the seven
had been renamed upstream. Nothing checking only the shape of a
reference could have found either.
The code and its tests went in hours before the record was touched, so
an issue that read `located` had been fixed all along. That is the exact
failure the frontmatter exists to prevent: status is meant to be
answerable from the record rather than by reading the code.
Closed with the commit that did it, and cross-referenced to 022 and 023,
which came out of the same mistaken instinct — treating the machine as
a boundary, then as an identity, then finding a consumer had a password
and no name to present with it.
The module descriptions sit in `examples/` inside the control plane, and
that name has been doing harm: everything there reads as a sketch, and
one shipped naming a container image nothing builds. A directory called
the catalogue would have made "does this work" the obvious question.
The shape of the answer turns on one measurement. Of the 126 modules in
the system being replaced, 47 are software in their own right — the
largest is 182 source files, and a speech-capture module carries a whole
daemon. Another 44 ship helper scripts. Only 35 are a description and
nothing else.
So a catalogue cannot be a folder of manifests, because two thirds of
modules are programs. That splits them four ways, and only two of the
four belong in a catalogue: things the world made that we describe, and
packages with some files. What the mesh is made of stays in the
repositories that build it. What we wrote keeps its description beside
its code, in the same commit, because nothing else can stop the two
drifting.
The mesh's list of modules is a table, not a repository, and it already
records where each module came from and at which commit. Nothing needs
inventing for modules from anywhere; a repository of ours is just the
source we curate.
The check that a description is valid should move to a command on the
control plane's binary. Today a test reaches into the control plane's
internals to parse manifests, and another reads its build file to check
images exist — two jobs tangled. A command would also give the same
check to somebody describing their own application, which is the case
that matters most and has none.
Left open: how a provisioner's image gets published and pinned, and
whether thirty-five install-a-package modules deserve to be modules at
all.
The cause was one line. Machines get a systemd-networkd unit with a
static address, so networkd finishes and reports the link configured.
The registry ran `ip addr add` inline, which leaves networkd waiting to
configure something it was never told about — and
systemd-networkd-wait-online has an infinite timeout.
So network-online.target was never reached and everything ordered after
it never started. On these machines that is Docker, so `docker load`
blocked on a socket whose daemon was queued behind a target that would
never come, and three bounded timeouts stacked to thirty-five minutes.
These machines have no DHCP by design, so that wait was never going to
end.
The hypothesis in this record was wrong and the record now says so.
Stocking had just been changed, so stocking looked guilty; stocking
takes 34 seconds and always did, timed directly before changing
anything.
Fixed with two things that made it cost hours instead of minutes: an
image placement now waits for the runtime and refuses after 120s naming
what systemd is waiting on, and the end-to-end test passes onProgress —
the raise reported every step and the test discarded it, which is why
thirty-five minutes and four minutes of silence looked the same.
The suite then ran to completion, 23 of 24, the one failure a check of
its own flagging a path as a credential because `/` is in the base64
alphabet.
Also recorded: a redirected log lags, because Node block-buffers stdout
to a file. Read as a stall twice, the second time right after the real
fix — where a buffering artifact argues the fix did not work.
Seen twice today. Once mid-run: thirteen passes, then the process ended
with no summary, no failure and no receipt. Once from the start: the
first test ran 35 minutes against a measured 4.5 and was still running
when it was stopped.
Ruled out rather than assumed: not memory (84 GiB free, no OOM), not the
daemon (the stalled machine answered `incus exec` immediately), and not
the changes under test — the anchor VM had no host log and no
containers, so the run never reached placing the host.
What changed just before is that the rebuild went from two artifacts to
six, and every one of them is pushed into the scenario's registry, which
is the step the second stall sat in. Recorded as what changed, not as
the diagnosis.
The reason this is an issue and not a slow test: the suite prints
nothing between starting a scenario and finishing its first test, so
four minutes and thirty-five look identical from outside, and the only
recourse is to guess. That is how a workstation was left unbootable in
August. And a run that ends silently after thirteen passes is a run
somebody may believe.
023 is fixed, so the design faults are gone and one concrete thing is
left: the realm provisioner does not exist. Its manifest named an image
nothing builds and no program backs, which has been removed — a manifest
describing a program nobody wrote is the same mistake as the credential
files that could never be read.
Keycloak's manifest now says what is true today, and the gap is loud: it
no longer claims to provide oidc-client, so a consumer asking for one is
refused by name at plan time instead of resolving cleanly and waiting
for a client nothing will create.
The provisioner should be written against a real Keycloak in the lab
rather than from the API documentation. The object store's took three
corrections that only a running server produced.
Tool servers — 56 modules, over half — were written up as the biggest
missing thing. They are expressible with what exists, and the first
framing was wrong in a way worth keeping: a module provides `tools` and
the session requires them does not work, because a requirement has one
answer and 56 modules offering tools would be 56 answers.
Turned around it fits exactly. The session provides `tool-host`; every
module offering tools requires it and contributes where its tools are.
Many-to-one is what `contributes` has always been, and the session
receives all of them in one file. Verified by resolving it rather than
by reading the code.
It only became possible today: until 022, several modules on one node
requiring the same thing was refused outright. Worth noting because it
means the credential fix bought more than credentials.
What remains is a decision about what a tool server is, which is work
rather than a missing shape.
The entry stays in the list rather than being deleted — a checklist that
quietly loses its biggest item reads as though nobody looked.
Both halves had one cause: the mesh knew something and did not say it.
Who a consumer is now comes from one derivation, sent to the provider in
its grant and to the consumer in its binding, so the two agree by
construction. The provisioners use the name they are given and refuse to
invent one, because a name of their own would create a login the
consumer could never guess while everything reported success.
Bound values reach the file that needs them through the symmetric twin
of the sealed placeholder — simpler, because they are not secret, so the
control plane fills them in and the host gains nothing.
The lab run meant to prove this failed in a way that looked like the fix
being wrong: rotation could not authenticate against a real database.
The cause was the suite rebuilding the control plane's image and not the
provisioner's, so an image built that minute ran against a provisioner
built the day before. That is 005's family and is recorded with the
issue, because the misleading part is worth more than the fix.
All three modules have manifests, all three parse, resolve and plan, and
none of them can start. Worth writing down before it reads as progress
or as failure, because it is neither.
The vocabulary held. Nothing in 3.1–3.3 needed a new shape — including
the mail system's several containers on a private network, which was the
one expected to break it. That was the question this phase was designed
to answer.
What did not hold was underneath: 022, now fixed, and 023, open. Both
are about credentials rather than about what a module can say.
The third fault was in the manifests, not the design: a secret declared
at a path named .env and read as one, when a sealed file holds a
password and nothing else. That is what a manifest checked only by a
parser buys, and it is why there are now two tests reading the manifests
on disk.
023 is the whole of what remains before the identity provider runs.
I wrote that a module cannot declare an action. It could — the parser
accepted one, and the refusal only came on the machine. The claim was
wrong in the direction that matters: it read as "the design prevents
this", when what prevented it was a check at the far end that nobody
would connect back to the manifest.
Health checks are still the gap most worth closing, but the shape of the
answer is different from what I wrote. An action is not available to a
module at all, so a health check needs a way to say ask this and expect
that without saying run this — closer to a listens entry than to an
action.
Also records the finding itself, because it is a recurring shape here
and not a one-off: a rule enforced only at the far end is enforced and
unusable.
The playbook offered `env-file` and `${secret:name}` as alternatives,
and that reading is what produced the bug every example module shipped
with: own-secrets pointing at a path named `.env`, mounted as env-file,
holding a bare password. The container starts with no password set —
which is a service running on the wrong credential, not a failure.
They are not alternatives. A sealed file holds a password and nothing
else, so env-file points at a file the module declares whose content
leaves a hole, and the host fills it on the machine. A provisioner is
the exception, because it reads a password file.
Written out as the three lines a module needs, with the failure it
prevents named, since the abstract version was already there and was
read the other way.
Amends the credentials page, which said "every pair has its own
credential" and meant two machines. Built that way, it was wrong in a
way that only shows on a real node: a machine running several services
against one database server had one credential between them, so the
provider refused to plan at all and the consuming node quietly gave the
first module a credential and the rest nothing.
The page already argues the case against itself — one credential with
many holders is the first of the three faults it was written to remove.
It just drew the boundary at the machine.
Two modules on one node are as separate as two on different nodes, and
one login opening both is what this page exists to prevent. It is also
what makes withdrawal possible: one role per machine cannot say that
this module has lost its login and the others still have theirs.
022 turned out to have a silent half worth recording: the provider
refuses loudly and names the modules, which reads as a decision, while
the consuming node does not refuse at all. Three modules wanting one
database produce one need, so two of them get no credential file and
each starts and fails to authenticate with nothing saying why.
023 is what remained after fixing it. A consumer now gets its own
password, in whatever shape its configuration wants, and still cannot
connect: the user name is invented by the provisioner and recorded
nowhere in the mesh, and the host and port sit in a JSON binding that an
application reading KEY=value cannot use.
The asymmetry is backwards and the coverage document now says so. The
secret is the hard case, because the mesh must not be able to read it,
and the secret is the part that arrives. The host and port are ordinary
facts the mesh holds in the clear, and they are the ones stuck.
Keycloak, Gitea, Mailu and MinIO all parse and resolve and none of them
can start. This is what stands between the module set and a running one.
Found while checking whether the module vocabulary covers real use
cases. A node running three modules that all want a database cannot be
planned at all:
anchor has 3 modules asking for "postgres-database" and they would
share one credential: gitea, keycloak, umami
The refusal is right about what it says and wrong about what it implies.
They would share one credential, and sharing is worse than refusing —
but the arrangement being refused is the ordinary one, and the node this
mesh exists to take over runs eight modules against one database server.
The cause is the key: a credential is keyed by provision, consumer node
and provider node, so `consumer` is a machine. The provisioner inherits
it and names the role `mesh_<node>`. The refusal is not a check that
caught something; it is the only honest thing that function can do with
a key that cannot tell two consumers apart.
It is the same mistake as 021 with a different face. There the machine
was treated as a trust boundary; here it is treated as an identity, as
though "who is asking" is answered by naming a host. Two modules on one
node are as separate as two on different nodes.
Worth stating plainly: without the refusal, gitea's login would have
opened keycloak's database, and nothing would have said so — from the
provisioner's side it created exactly what it was asked to create.
Not a local fix. It crosses the control plane, the grant file naming and
every provisioner that names something after a consumer.
Every manifest in the system being replaced was read and every key
counted, then set against what the new one can express. Three findings
worth more than the table.
**The most-used key was already covered and I expected a gap.**
Depending on another module — 65 manifests, the commonest thing any of
them says — is a requirement naming a module, which already means that
module rather than anything providing the name.
**The largest real gap is tool servers: 56 modules, over half.** A
module can already run one; what is missing is anything saying it offers
tools. That is plausibly a provision rather than new vocabulary, which
would need nothing added — not yet decided, and recorded as undecided.
**The gap most worth closing is health, at seven modules.** The mesh
knows a container is running, which is not whether it answers, and this
project has paid for that distinction twice. An action with a verify is
exactly the right shape and may not arrive over the link, so a module
cannot declare one.
Two things are missing deliberately and say so: stage hooks, because the
link may not carry an action and a module needing setup ships a program;
and flavours, retired in favour of claims.
Config merging is missing and should stay missing. A mechanism that
understands TOML gets asked for YAML, then INI, which is how the thing
being replaced became unholdable.
Also records what the survey found that is not about coverage: manifests
that had stopped matching what was actually brokered, one fact derived
in two places giving two answers, and a live listing returning
credentials in plaintext.
Written after porting the first real workload end to end. Every step
exists because skipping it cost something, and the ratio is recorded
because it is the lesson: six attempts, one real bug, and the mesh was
right every time.
The rule worth carrying out of it: read the host's log before
theorising. A declaration that was sent and not applied says so there
and nowhere else — it took an hour to look, and the answer was one line.
The conversion's detail is operational and names machines, so it lives
in the mesh's knowledge base rather than in this repository:
`migration/where-service-data-lives` for where every service's data
actually sits, and `troubleshooting/db-password-frozen-at-first-init`
for the lockout. This document says the rule; those say the specifics.
The lockout is the finding worth carrying here, because it is worse than
the one this plan was already guarding against and it is likelier. A
database image consumes its password variable only when its data
directory is empty. Everything keeps data on a persistent directory, so
the role holds whatever password it was created with for ever;
regenerate the variable and the application moves on while the database
does not, permanently, because nothing reconciles it.
Eight modules are in that state today and work only because nobody has
regenerated their credential since their data directory was created.
It was already documented in the knowledge base and my survey had missed
it — found by searching, which is the argument for the knowledge base
existing.
Pinned 2.5.0 rather than latest, on the suspicion that its draft
profiles extension was involved. Identical failure, so that is ruled out
and recorded — two of the three guesses in this issue have now been
tested and both were wrong, which is the useful half.
The scenario keeps the pin regardless; it should have had one from the
start.
Against a real ACME server the proxy orders, the challenge is answered
at the name on port 80 through the proxy itself, the authorisation goes
valid, finalisation is accepted, and the authority issues a certificate.
The client then posts to an empty URL to collect it, and never does.
Read from the authority's own log rather than inferred. Across one run
it issued two certificates and accepted finalise three times: the client
reaches issuance every attempt and fails at the same step after it.
Ruled out and recorded, so nobody repeats it: the directory is complete;
the authority's API certificate covers the address; the challenge path
works. A hand-written server config was suspected and was wrong —
replacing it with the server's own default, changing only the challenge
port, gives the identical error.
Filed rather than pursued because what remains is interop between two
libraries against a server that exists to be a test server, and may say
nothing about a real authority. What the mesh needed to show, it showed:
a routed name gets a certificate ordered from a configured authority,
and an unrouted one gets nothing — that second assertion passes.
Phase 1 closes with this one item partly open. Two of its four tasks
needed no code at all, the network shape was built, and the next thing
to learn comes from moving a module rather than a fourth lab run.
An earlier paragraph implied a secret becomes unrecoverable once
accepted. It does not. It is sealed to the node, which holds the private
half and writes the plaintext into the module's own file at 0600 — the
value is there, on the machine, as an ordinary file.
What does not exist is a way to ask the mesh what a secret is. That is
the property worth having and it is narrower than what was written.
The reason to capture the old system's environment first is simply that
adoption means supplying those values, not that they become
unrecoverable.
Nothing is rotated during the conversion. A service keeps the password
it is already using, because minting a new one is how a running service
stops being able to reach its own database mid-migration.
The mesh has both paths already: generate-and-seal for a new module,
accept-and-seal for an adopted one. Adoption needs the second, and it is
built.
Rotation becomes a separate act afterwards, once everything works — the
machinery is proven, and it is a thing to do deliberately rather than as
a side effect of moving a service between systems.
Records the step that has to come first and is easy to miss: read the
current environment out of the old system while it can still be read.
Once accepted, the mesh cannot show a secret back, and once the old
system is gone neither can that. A password nobody wrote down is a
service nobody can adopt.
Assumed throughout and stated nowhere — the wrong way round for the most
consequential fact about this component.
A board reachable only over the private network would sit inside the
boundary 0004 already calls the security boundary, and a login there
would guard a room whose door is inside the building. This one faces the
internet, so its login is a perimeter rather than defence in depth.
Which makes the identity provider the mesh's outermost gate. The board
presents the control plane, and the control plane's networked surfaces
can change the mesh (0035) — so whoever that provider admits can assign
modules, from anywhere. Written flatly because it is easy to arrive at
one reasonable step at a time and then be surprised by.
What follows is not the board's own design: who may log in is a decision
about the mesh rather than about an application; a public name needs a
certificate from an authority the world trusts, which is why that work
exists; and the provider going wrong in the permissive direction is a
mesh-wide exposure with no local symptom.
The command line is unaffected and is why this is tolerable — it
authenticates through nothing and answers to the machine's own login, so
the mesh stays operable by somebody standing at it whatever happens to
the gate. That is the property to protect if the rest is ever traded
away.
Bootstrap stopped when the control plane started — a mesh that runs and
cannot be used by anybody not standing at the machine, since the
networked surfaces need an identity provider and no module has been
assigned yet. It now runs through the provider and the first login.
The obstacle was not incidental. The mesh has never held a readable
secret: Make generates and seals, keeping no readable copy. An initial
administrator's credential is the first value a person must read.
Generating it and printing it once was the convenient option and is
refused. It would give the control plane a plaintext secret for the
first time — briefly, and to one terminal, but the capability would then
exist, and an exception made for one case does not stay one. The next
awkward credential gets printed too, and "a copy of the database is a
copy of nothing" stops being checkable by reading the code.
So the operator supplies it, on standard input, not echoed — the path
that already exists for a model-access key. What is created is an
account in the identity provider, not a user of the mesh; there is still
no user model.
Unattended bootstrap remains possible and the value still comes from
outside: automation supplying it is the operator supplying it. What is
refused is the mesh inventing one, so an unattended bootstrap with
nothing provided yields a mesh with no administrator — correct rather
than broken.
The mesh is operated from a command line and must be operable from a
browser and from a model's tools, without becoming three systems. The
pattern is already in the code and was unnamed: `board` serves HTTP by
calling the same functions the CLI calls, holding nothing.
Takes the decision 0034 said had to be taken deliberately rather than
arrive with a feature: the HTTP surface is not read-only, so a browser
login now carries authority over the mesh.
Names the dependency by protocol — an OAuth2 identity provider — as the
mesh does for AMQP, S3 and OCI. Keycloak is what fills the role; what
the control plane knows is that it validates a token, and replacing the
provider is a migration rather than a redesign.
Says what this must not become, because it is the failure the project
was started over: a kernel every module imports, 155 files of code from
every context. Shared surfaces are not a shared library. Three adapters
calling the same functions is not the same as logic leaving the context
that owns it.
And records the loop it creates. The networked surfaces depend on a
module the control plane assigns, so when identity is down nobody can
authenticate — including whoever is trying to fix it. The way out is the
command line, which authenticates through nothing and is available to
the account that owns the machine. Hence the rule: no capability exists
only behind an authenticated surface, because that is a capability which
disappears exactly when identity does.
Supersedes 0032, which decided the right thing and described it wrongly.
The decision is unchanged: the account that installed the host owns the
mesh, and there is no user model.
What was wrong was inventing "a surface that delegates authentication"
for the board. It is a web application with a login, in the way every
web application has a login. That is a fact about an application, not a
property of the mesh.
The cost was not cosmetic. It made the identity module look like part of
the mesh's authority — something the mesh depends on to know who anybody
is — when the mesh knows nothing about people at all and one of the
applications running on it happens to have a login.
Keeps the line that is worth writing down, and states it more plainly:
signing in to an application must not become authority over the mesh.
Today it cannot, because the board reads and does not act. The moment it
can assign a module, whoever it lets in has mesh authority — and it
would arrive as a feature rather than as a decision. So a surface that
can change the mesh is a change to who owns the mesh, and is taken as
one. Not forbidden; just not something that turns up in a pull request
titled "add assign button".
Third correction to one table today, found the same way as the other
two: by asking whether both halves of the test were answered, or only
the easy one.
0006 admits the registry because "it cannot grant itself a repository" —
true, and the second half. Nothing established that the control plane
needs one in order to run. Counted rather than argued: the bundle raises
twelve resources and no registry is among them. The registry arrives
afterwards as an ordinary module, which is exactly what the lab asserts.
0006 half-said this already, calling it "substrate by role and ordinary
by delivery, provisioned once there is a control plane to do it". A
member provisioned by the thing it supposedly precedes is not a member;
that phrase was carrying a contradiction rather than resolving one.
The registry is a closer call than the object store and the difference
is worth keeping: the control plane never touches an object store at
all, but it genuinely uses the registry. So the registry is a real
dependency of the mesh operating and not of the control plane starting —
and it is the second that the word means.
The substrate is now exactly what the bundle raises, which is the
strongest form the list can take: checkable by counting rather than by
reading an argument, and the two cannot drift.
The finding is not about substrates. A test with two conditions is a
test only when both are asked.
Answers what 0031 left open, and a question it did not ask — who owns
the mesh at all. There was no answer, and the absence was invisible
because every operation so far has been run by the person sitting at the
machine, so nothing had to say whether that was the design or the
circumstance.
The account that installed the host owns the mesh on that node. No user
model, no roles, nothing to administer. It follows from 0004 rather than
adding to it: there is no authorisation between nodes because every node
is the operator's own, so a user model inside that boundary would guard
nothing — anyone it could stop could read the node's key off the disk.
The board is different, and the difference is the network. A surface
reachable by a browser has to know who is asking, because those people
are not by construction people with a shell on the machine. So it
delegates to an OAuth provider, which is a module.
That does not make identity substrate. A surface delegating
authentication is not the control plane delegating it: the control plane
runs, applies declarations and reaches nodes with no identity provider
in existence. Only the board needs one.
Records the cost plainly: anybody with a shell on a node has full
authority there, and there is no way to give somebody authority over one
node without giving them a login on it.
Closes the last open question about what the substrate contains. 0006
left an identity provider conditional — substrate only if the control
plane delegated authentication — and said the decision had not been
taken. It is now: it delegates to nothing.
The conditional was never about machines. A node proves itself with a
keypair it generated over a broker account issued at enrolment, and
declarations are verified by signature; none of that involves an
identity provider. It was only ever about whether a person signing in to
a mesh surface would be authenticated by something else.
So the substrate is three — a relational store, a message bus, an image
registry — and with 0028 having removed the object store, no member is
conditional and every one is there for the same reason.
It does not settle how a person signs in to a surface, deliberately.
What is settled is that whatever answers that is not something which
must exist before the mesh does, so it can be decided late or replaced —
which being substrate would have prevented.
The conversion method, recorded because it decides everything else and
was not written down.
The old control plane is stopped — provisioning, coordinator, syncs, the
pipeline, anything that decides or writes. The workloads it was managing
keep running, because nothing is managing them. The new mesh then takes
ownership one module at a time.
Nothing is ever unassigned in the old system. Unassigning is how it
removes things and removing is how data is lost; it is asked to stop
having opinions, never to take anything away.
Disabled rather than merely stopped, which is the part easy to get
wrong: those units are enabled, so a stop lasts until the next reboot. A
reboot mid-conversion would bring the old control plane back to
regenerate managed files underneath the new one — the one situation
where two systems really would fight over a machine.
A service left running with nothing managing it is the safe state: it
has its data, its configuration is on disk, and nothing will change
either. The risk in a conversion is in the managing, not the running.
Also records why taking ownership piecemeal is safe: the new host's
orphan removal is per-origin, so it only removes what it recorded
itself. Services it was never told about are not orphans to it.
0030, found by asking what the conversion actually needs rather than by
reviewing anything. The host deleted a directory and everything under it
when it stopped being declared — which happens when a module is
unassigned, or when a manifest is edited to move a data folder, which is
the exact operation this plan needs. A database's files, a mail spool.
The report said "removed".
A directory still holding something is now kept and said so. No flag and
nothing to remember: emptiness is the test, and it works because the
removal order was already right — the mesh's own contents are gone by
the time the directory is reached, so what remains is by definition
something nobody declared.
The plan now says data outranks its own ordering: copy, read back
through the service that owns it, and only then point anything at the
new location. Never move and then check.
And it records where this starts — the node holding all the production
data — with what that costs stated rather than argued with. Everything
proven so far was proven on machines that could be destroyed and raised
again. A scenario proves the mechanism, not the state on that machine.
Recorded because it is load-bearing and was not written down: moving
from the current system to this one is a person at a command line, not a
migration program.
What that removes is larger than what it adds. Nothing in this plan
needs an importer, a translation layer, a compatibility shim, or a way
of keeping two systems agreeing while both are live — each of which
somebody would otherwise reasonably build, use once, and maintain for a
year.
It also settles what "safe" means for the system being retired: a fix to
it must be safe on its own, because there is no careful rollout to
sequence it into. A change needing three steps in the right order is a
change that will be half-applied. That reversed a certificate default I
had chosen this morning.
Ordering needed no change for the third time running — resources apply
in the order declared and nothing sorts them — and is now asserted,
because sorting them for any sensible reason would have passed every
other test.
Separates ordering from readiness, which the task had run together: a
container started is not a container ready. Nothing waits, and what
needs something usable retries. That is deliberate and more robust than
start ordering, since a dependency can restart long after apply.
The network was the first thing in Phase 1 that genuinely needed
building, and the first that needed a decision: 0029 records why a shape
rather than an action, and the vocabulary is nine.
A session as a licence consumer needed no change either: the two
sessions are two modules, so the existing (node, module) binding already
names them apart. 14-model-access.md's "a step toward it and not it" is
true of a worker and not of a session, and the difference is that there
is one session per node rather than many per machine.
Records what stays open: the worker half of that gap is real and
unaffected, and belongs with 0003, which is unbuilt.
Two tasks in a row that were already possible. Both were written from
the design rather than from the code — the review's own finding arriving
in the plan it produced. The remaining Phase 1 items should be checked
against the code before being started rather than after.
An object-store provision, proven against a real store with seven
assertions.
The finding is worth more than the task: the control plane
special-cases nothing. provides, requires, contributes and grants are
name-agnostic, so asking for a bucket needed no change to the mesh at
all. What was missing was a provider and the last step on the machine —
"add an object-store provision" was never mesh work, and the breakdown
now says so rather than leaving the next person to rediscover it.
Named s3-bucket by 0027: the coupling is to the API, not the product,
because swapping one store for another does not break a consumer. A
database is the other case and names its engine.
Records the assertion a database does not need, because it is the one
that will be forgotten when somebody writes the next provider: one store
holds every bucket behind one endpoint, so isolation is a policy rather
than a property, and a policy granting everything passes every test that
only checks a consumer can reach its own bucket.
**0024 accepted.** Model access was decided, built, and proven in the
lab, and two design documents rest on it; only the status had never
moved. The gate is green again.
**The work breakdown rewritten.** It planned a decomposition of the
existing system in place — extract contexts, declared features, shrink
the shared library. That is not the work. A replacement is being built
beside it, and only the old Phase 0 survived contact with reality, so
the one document meant to say what happens next was describing a system
being retired.
Now ordered by what "modules move across one at a time until the old
registry is off" actually requires:
- Phase 0 is marked done against the twenty-two lab assertions, **and
carries its own limitation**: every module exercised was written to
test the mechanism, so the vocabulary was shaped by its own fixtures.
- Phase 1 is the vocabulary gaps found by asking what real modules
need — an object-store provision, a session as a licence consumer, a
network shape with ordering, public certificate issuance.
- Phase 2 is one module, then a week of running it, because the point of
going first is to find what Phase 1 missed.
- Phase 3 picks modules that each prove something the first did not; the
mail system is last because it is the one that may send work back into
the declaration language.
- Phase 4 is switching the registry off, named as a phase so it is not
mistaken for the goal.
Keeps the rules of engagement unchanged — they were about how work is
done, not what it is — with one addition: stop and ask before anything
that touches a machine outside the lab.
Adds a section on keeping the list true, since the document it replaces
was wrong for weeks and nothing said so. A claim here is counted, not
reasoned, and a phase is done when the lab says so.
First pass of a design review, done by reading documents against code
and against a raised mesh rather than against each other. Every error
below was invisible to a proofread.
**Statuses were stale, and nothing checked them.** Ten to-be documents
said `designed` while naming working, lab-proven code — several with a
*What was built* or *Raised, and observed* section. Added a
`status-vs-code` check: naming a file is a claim that the file
implements this, so a document that points at one has stopped being
merely designed. It failed on all ten before it passed, per the rule
this folder sets for its own checks.
**The bundle carries three images, not two.** 07 reasoned about which
substrate services go in and overlooked that the control plane is in
there too — it is what the substrate exists to start, and there is
nothing to fetch it with yet. Counted, not deduced.
**The bootstrap uses four shapes, not six.** It listed `file` and
`directory`, which substrate-first-node.lock never asks for. The claim
that mattered — nothing is blocked on the host — was true either way,
which is why the wrong count survived.
**The eight capabilities were documented nowhere.** Implemented in
internal/profile/detectors.go and enumerated in no document, including
the one about the host that detects them. A vocabulary modules write
against, readable only by reading the code. Now written down, with the
seat/graphical-session distinction that is wrong in both directions if
collapsed.
**MinIO swept out of the to-be layer** per 0028.
The gate now fails on one thing left deliberately: ADR 0024 is
`proposed` while two documents rest on it and the feature it decides is
built and lab-proven. Accepting a decision is not mine to do.
**0027 — provisions.** A module written against PostgreSQL could be
matched to a provider of SQL Server, resolve as satisfied, and fail on
its first query. The name said the role, so nothing distinguished
engines. Refusing on ambiguity could not help: with one provider of
each name nothing is ambiguous. Enforced at parse rather than
documented, because the old naming was the documentation.
**0028 — the substrate.** 0006 admits an object store on the grounds
that it cannot grant itself a bucket. That answers the second half of
the test and assumes the first: the control plane does not need one.
Verified — no S3 client in mesh-control, and internal/builder/registry.go
records the deliberate choice to put artifacts in the OCI registry as
content-addressed blobs. The row was inherited from the system being
replaced, where an object store distributed module tarballs, and was
never re-tested against the definition above it.
So an object store is an ordinary module, and a mesh with nothing
needing one runs none. Migrating it is module work, not substrate work.
0028 also states what 0006 left unsaid: a substrate service and a
module of the same product are different instances. The substrate is
raised from the bundle before any mesh exists, so it is not in the
module graph — a workload depending on it would depend on something the
graph cannot see, cannot rotate a credential for, and cannot move, and
would put workload data in the store the control plane keeps its own
state in.
Both records were found by reading code against design rather than
design against itself, which is the review that should have happened
sooner.
Answers the question 15 raised: a board showing many sessions leaves
one-per-node untouched, because each is still one conversation. Only
concurrent conversations with the same session would touch 0004.
Records soulstream and herdr as the prior art to draw from, and marks
it explicitly off the provisioning path so it stays a note rather than
becoming the work.
Settles the question 15 left open: the mesh session holds its own
memory in the mesh root, rather than assembling a view over the node
sessions. Memory follows the rule the rest of the design already uses —
the context root is the whole of what makes one session a different
agent, and memory is part of what makes it that agent.
The control-plane node is what makes this load-bearing rather than
tidy. Two sessions share that machine; if memory belonged to the
machine instead of the root they would share it too, and the mesh's
recollection would be indistinguishable from that node's own — the
collision 0026 exists to avoid, arriving through the back door.
Also corrects an error made writing it up: memory is NOT declared
state. The engram and tools are — the mesh says what they are and the
host writes them (0011). Memory is written by the session itself and
declared by nobody, so a mechanism that regenerates the root wholesale
would erase it on the next heartbeat, silently, while reporting
success. The root is not uniformly managed and which parts are has to
be explicit.
A session for the mesh itself, addressed as the mesh, differing from a
node's in exactly three things: the context it starts in, its engram,
and its licence binding. Not a new kind of agent — the same mechanism
pointed at a different root. Two implementations of one mechanism drift,
and the vocabulary collision 0001 exists to undo began exactly that way.
It runs on the control-plane node, and the reasoning is easy to get
backwards: not "the important agent on the important machine", but that
this node is already the one place excepted from "compromise of a node
is compromise of that node". Placed anywhere else it would create a
second such place.
It is an addition to per-node messaging and never a replacement. 0001
holds that losing the control plane costs change, not operation — and a
mesh whose only conversational surface lived there would lose the
ability to ask anything while every machine kept running perfectly.
Writing it up exposed that the node session's setup was never designed
at all. 0004 gives behaviour and stops: nothing said how a session
starts, where its context lives, or how a broker message becomes a
prompt. That gap was invisible until something had to be built *like* a
node session. 15-the-agent-session.md covers both as one mechanism.
It also makes "a consumer that is not a machine" undeferrable. The
control-plane node now hosts two sessions that must hold different
licences, and a per-machine binding cannot express that at all. Noted in
14-model-access.md against the gap it was already recorded as.
Also completes the to-be index, which stopped at 10 and omitted four
documents. Pre-existing broken ADR references in the older rows are left
alone rather than guessed at.
Decides the question 006 narrowed to. An agent reads this repository
directly and the search consults it, so these documents surface beside
ordinary results instead of only when somebody already suspects they
exist.
A scheduled sync into the mesh's memory was the option that works with
what exists today, and lost on the ground this repository can least
afford: it makes a second copy, and the copy that is searched quietly
stops matching the copy that is edited. A design record that has
silently diverged from the reasoning it claims to carry is worse than
one that cannot be found — the first misleads, the second merely fails.
Amends what 0019 promised rather than satisfying it: these documents
will not be indexed, they will be read. The commitment that survives is
the one that mattered — that a searcher finds them without already
suspecting they exist.
Gated on an agent that does not exist yet, so 006 stays open on the
build with a decided shape. What closes it is a check that fails today
by design: search the mesh's memory for a phrase that appears only in a
design document here, and require it back.
**004 — certificate issuance.** The resolver declared no authority at
all, so the client fell to its built-in production default: there was no
setting set wrongly, there was no setting. It is now a node property
defaulting to staging, which answers the first open question. Staging by
default rather than production-with-an-override, because the alternative
leaves the safe path depending on remembering to opt out of it — 005's
lesson, in a second place. The rollout is ordered and the order is the
dangerous part; recorded, not performed.
**008 — node rescue.** Read back from running nodes as the report asked,
and one of its own claims was wrong in a way that matters: the health
timer does exist and does fire. It simply never calls the rescue script.
A trigger that exists and does not do what the script claims survives a
halfway check, which makes it worse than the absence the report
described. Resolved by making the documentation true, not by
implementing rescue — the replacement host already supervises recovery,
and wiring unattended restart into the fleet being retired is a
deliberate decision rather than a tidy-up. Two "self-healing" claims
narrowed to what they actually do.
**006 — deliberately not closed.** Re-checked today: the indexing still
does not exist. What is gone is the reason it was an issue — the claim
is no longer load-bearing, because the README names the gap and the
decision's reasoning never invoked indexing. A signpost now points here
from the knowledge base, and was measured rather than assumed: it is
reachable, it is not surfacing. Closing it while the indexing does not
exist would be this repository's own named failure, one folder from
where it names it.
Retired in favour of the lab rather than repaired — that answers the
first open question. The second finding is the one that generalises:
"nothing runs it, and nothing reports that nothing runs it" is not a
fact about that harness, it is a fact about any suite too expensive to
run on every push. The replacement inherited the fault it was replacing.
Records the three rules that now hold, and what the fix taught twice:
the remedy rebuilt the symptom inside itself, and the code that counts
results passed every test while reading nothing.
A service is reached at <service>.<node>.internal, so what resolves is anything
under a node's name. The mesh writes the data and runs no daemon; two roles,
two claims, because systemd-resolved cannot serve a wildcard at all.
Both prohibitions were found by a machine rather than by reasoning: an address
systemd already held, and reading resolv.conf for upstreams that now point at
itself.
Twice in one file, a statement about a machine that read as reasoned and was
wrong — and the module's unit tests all passed while the daemon could not
start. That is what a unit test is: it confirms the assertion was made, never
that it is true of any machine.
003 in prose rather than in a manifest key.
The field was called needs, beside secrets, and both were name-to-path holding
something secret. What separates them is whose, not how secret — so that is
what the name says now.
012 named its own closing condition — a scenario with four images coming up —
and the scenario now stocks seven and has raised cleanly many times at the
memory the wrong diagnosis had raised.
001 is answered by the host reading the package database back after installing.
002 was NOT answered and was present here too, so it is a fix rather than a
note: a stale index is now named instead of reported as a failed install.
The connectivity design still said a hub cannot be filtered — a gap recorded in
the morning and closed in the afternoon, left standing as though it were
current. Worse than a stale date: it would send somebody away from something
that works.
`restart-on` was described nowhere, including the part added today that lets a
service reflect a file another module put on the machine. A rule the host
enforces and no document mentions is a rule nobody can rely on.
And nine of fifteen design documents claimed an `updated:` older than their last
change, some by a week. That field is what cross-cutting views are generated
from, so it is not decoration.
A resolver takes over /etc/resolv.conf, which is a singular resource — ADR 0009
lists it in the table beside the seat and pid 1. So choosing between resolved,
dnsmasq and unbound is assigning a module, per machine, and the mesh refuses
two rather than letting them fight over the file.
Recorded because it was treated as an open question two days after being
decided, which is the argument for that table being a table.
Found by a container failing to resolve a name every machine could: a container
gets its own hosts file holding only its own hostname, and on the machine it
always worked, which is what made it easy to miss.
Declared containers are given the names. A container somebody starts by hand is
not the mesh's to configure — which is a second, different reason to want a
resolver, recorded beside the first rather than folded into it.
Asked whether a machine that drops off needs re-adopting: it does not, nothing
expires, and the only thing that forces re-enrolment is losing its own key.
The gap was the twenty or thirty seconds after a resume in which a node
believes it is in a mesh it has left — recovering on its own, which made it a
quality gap rather than a fault, and still a machine waiting to be told
something it already knew.
It meant failed-or-refused, so the question this record says must not be lost
was answerable only for the machines that broke. Out of date, never told, and
not worked out are kept apart: the remedy is the same push and they read
differently to whoever is looking.
Their subject matter has been built and proven for days and their frontmatter
still said code: [] — which is what the cross-cutting view is generated from,
so it was claiming nothing existed for the substrate, the node lifecycle and
delivery.
One reading answered three ways, holding nothing and touching no context's
store — which is the constraint the whole document is about, and the thing the
board being replaced gets wrong.
Rotation and the provisioner contract; model access as a provision answered by
a record, with ADR 0024's other two gaps left as gaps; exposure, which closes
the open question about revoking a route; and the delivery loop, which closes
the gap ADR 0010 left when it replaced a pipeline with a comparison.
The broker's fingerprint travels with its credential, and the machine's
filesystem does not travel at all — it runs in a container, which is the
arrangement working rather than a limitation to route around.
Found by the firewall: every packet filtered as declared, and the machine
reported as not doing what it was told, because the unit that loaded the rules
had finished. Stated as a gap rather than worked around silently.
Otherwise it succeeds into a state its verify rejects, and the host's report is
accurate and names nothing. Recorded where the vocabulary is described, because
it is a rule about writing an action rather than about one action.
A hub needs its overlay port open and a node that is not a hub does not, and
they are the same module — so listens, a static manifest field, cannot express
it while the overlay module's resources are computed per node. Written down
rather than left as an oversight for whoever first puts a firewall on a hub.
Issue 014: the node's serving key was stored in the host's own encoding, so
every check that reads the file passed and no server could start. Same shape as
013 — two halves of one mechanism designed separately, each correct about its
own half. Where a file exists so a third party can read it, the format is the
interface.
Issue 003 is answered in both halves: manifests are parsed strictly, and a
module says what it listens on and from where rather than carrying a key
nothing reads. The design records what was built and how each part is checked.
Issue 013 is new, found by reading while writing the first module that has
both a computed file and a service that needs it. The file arrived second.
It failed, then the next reconcile fixed it, which is why nothing caught it.
2026-08-31 00:37:34 +02:00
642 changed files with 49382 additions and 8938 deletions
description:Use when a thought is raised that should be remembered but NOT worked on now — an aside during other work, a "we should look at X someday", a known gap nobody is assigning yet. Triggers on "defer this", "park this", "register this thought", "note this for later", "don't work on it, just remember it". Records and returns to whatever was already in progress.
---
# hq-defer
Parks a thought so it is not lost, **without moving the work off course.** The defer is the
point: the thought is recorded and the previous task resumes.
This skill wraps no playbook, because deferring is not part of the development cycle — it is
what happens *before* something enters it. A parked thought has no number, no owner and no
status, and that is correct.
## Where it goes, and why not the repository
Record it as **one memory file** in Claude's persistent memory directory for this project (the
path is given in the session's memory instructions), with `metadata.type: project`, plus a
one-line pointer in `MEMORY.md`.
**Not** in `04-ISSUES`, the knowledge base, or anywhere else in the repository:
- A parked thought is not an issue or a research effort. Giving it a number asserts it has been
triaged, which is exactly what deferring says has not happened.
- A shared "deferred" or "someday" document is a **central status file**, which
[`AGENTS.md`](../../../AGENTS.md) forbids. Status lives in frontmatter on real records, and a
parked thought has no real record yet.
- A repository write means a branch, a commit and a pull request — drift, which is the one thing
this skill exists to avoid.
## Steps
1. Write the memory file. Slug is kebab-case and descriptive of the thought, not of the act of
deferring.
```markdown
---
name: <kebab-slug>
description: Deferred note — <one line>
metadata:
type: project
---
Raised and deliberately deferred on YYYY-MM-DD: **<the thought, in the user's own terms>**
**Why:** what was being worked on when it came up, and that deferring was intentional so
that work was not pulled off course.
**How to apply:** treat as an open thread, not an assignment. Do not start on it
unprompted. If it graduates it needs an HQ home first — an issue under playbook
[03](../../../00-META/process/03-issues.md) if a stated behaviour does not happen, or
research under playbook [01](../../../00-META/process/01-research.md) if it is still an
idea. Say which is undetermined, if it is.
```
2. Append one line to `MEMORY.md`: `- [<Title>](<kebab-slug>.md) — deferred YYYY-MM-DD; parked, no HQ
record, do not start unprompted`.
3. Convert relative dates to absolute before writing. "Last week" is worthless in six months.
4. Check for an existing memory covering the same thought and update it instead of adding a
duplicate.
## Then stop
Reply in **at most two lines** — what was recorded, and that it is parked — and **return to
whatever was in progress before.** Do not summarise the parked thought back at length, do not
propose a plan for it, do not ask which playbook it belongs to, and do not open anything.
If nothing was in progress, say only that it is recorded.
## Do not
- Do not create an issue, a research effort, a decision record or a design document.
- Do not create a branch, commit or pull request.
- Do not start investigating the thought, however cheap the first check looks.
- Do not name nodes, domains, addresses, absolute paths or usernames in the memory file — the
thought may later be quoted into this repository, which is public.
- Do not decide whether it is an issue or research when the evidence does not say. Recording
"undetermined" is the honest outcome and costs nothing later.
| Research | the knowledge base, not a file: `knowledge-base.ask`, and `knowledge-base.get` with `kind: research` and a number | `status` (`active` / `graduated` / `abandoned`), `date`, links (`became`, `touches`, `cites`) |
| Decisions | the knowledge base, not a file: `knowledge-base.holds` with a topic (the clauses in force, with their records), `knowledge-base.get` with a number, `knowledge-base.links` | `status` (`proposed` / `accepted` / `superseded`), `topic`, `date`, `deciders`, each clause's status, links both ways, what changed it since |
| Issues | the issue tracker, not a file: `mesh_call issue-tracker.list` (optionally `status`, `owner`, `phrase`) and `issue-tracker.get` | `status` (`open` / `diagnosing` / `located` /`fixing` /`resolved` / `wontfix`), `worker` and `stale` while `diagnosing` or `fixing`, `opened`, `located-in`, `fixed-by`, `amended-design`, `record` (the folder here, when it has one) |
## Steps
1. Read the frontmatter block from every file above — a grep across each tree is enough, no
need to load bodies.
2. Render what was asked as Markdown tables. Group and sort sensibly. The **ADR index** is one
of these views: records ordered by number, with title, date and status, and reconstructed
ones marked.
1. Read the designs' frontmatter (a grep across `03-DESIGN/` is enough, no need to load bodies), the
issues from the tracker, and the records from the knowledge base. An issue report carries no status;
where an issue stands is only in the tracker (ADR 0275). A decision record's or research effort's
status is only in the knowledge base (ADR 0290).
2. Render what was asked as Markdown tables. Group and sort sensibly. The **ADR index** by topic is
`knowledge-base.holds` with each topic of the reading order `02-DECISIONS/README.md` writes, in that
order, with a `bound` large enough for the topic (it says what it cut); it is the only place the list is
shown (ADR 0248). The knowledge base has no verb that lists every record by number: a view by number
is built from `get` per number, and says which numbers it read.
3. Flag anything inconsistent at the end, as flags — do not silently correct the render:
- a `graduated` or `abandoned` effort with an empty`became:`
- a `graduated` or `abandoned` effort with no`became` link
- an `implemented` design with an empty `code:`
- a `superseded` record with no`superseded-by:`
- a `superseded` record nothing`supersedes`
- an `open` or `diagnosing` issue with no activity
- a to-be design whose `decisions:`points at a record that does not exist
- a to-be design whose `decisions:`names a record `knowledge-base.get` does not find, or one that is
not `accepted`
- an as-is design whose claims are older than its `updated:` date suggests
## Do not
- **Do not write a status file.** Central status files are explicitly rejected. The view is
always generated, always ephemeral. This includes the ADR index — the hand-written one had
already drifted after a single addition, which is why it was removed.
- Do not infer status from prose. Trust only the frontmatter; if it is wrong, flag it.
already drifted after a single addition, which is why it was removed, and the generated one
written into `02-DECISIONS/README.md` made every two open decisions conflict (issue 297), which
is why that went too. `index.py` fails if the list is stored there again.
- Do not infer status from prose. Trust only a design's frontmatter, the tracker for an issue and the
knowledge base for a record; if one is wrong, flag it.
@@ -9,6 +9,7 @@ works — plus the engineering practice that holds across everything Novox build
| [`context.md`](context.md) | The environment — conditions, not aspirations |
| [`effect.md`](effect.md) | What is different when the work is done |
| [`how-we-build.md`](how-we-build.md) | The rules that hold across the mesh, each one earned. **The source of the mesh constitution** — the governed page the mesh injects into design sessions is derived from it. |
| [`glossary.md`](glossary.md) | One name per thing — the authority on vocabulary, and the words that were retired |
| [`repos.md`](repos.md) | Where implementation lives, and what each repository owns |
| [`process/`](process/) | The playbooks — how work moves through this repository, for engineers and agents alike |
@@ -34,7 +35,7 @@ comparing it against the code rather than by anyone noticing:
- It described the pipeline as having a separate builder process and a build stage that
packages. Neither was true after 2026-08-04; the documents stayed stale until 2026-08-06
([ADR 0010](../02-DECISIONS/0010-delivery.md)).
(ADR 0010).
- It listed the mesh as spanning a fixed number of named machines, which is exactly the
content this repository cannot carry.
@@ -44,10 +45,10 @@ symlinks at all — the rule is not merely "only the installer may link", and a
elevating linking to a principle points the opposite way from where this is going.
What exists today is that the installer owns and reconciles every link
([ADR 0012](../02-DECISIONS/0012-the-mesh-creates-no-symlinks.md)) — an as-is fact, recorded in
python3 00-META/checks/index.py the reading order in 02-DECISIONS/README.md is current
python3 00-META/checks/index.py --write regenerate it
python3 00-META/checks/records.py structure: links, no record written here, citations in the form "ADR NNNN"
python3 00-META/checks/index.py the reading order's topics are written, and no list of records is stored
python3 00-META/checks/cycle.py the development cycle: a design's layer, status, decisions and code; an issue report's frontmatter
python3 00-META/checks/words.py the glossary's retired words are not used, no daemon's line is written, and no word is defined twice
python3 00-META/checks/words.py --list tools the words the catalogue's copy must list
python3 00-META/checks/words.py --list lines the daemons' lines the catalogue's copy must list
```
Non-zero exit on any problem, so it can be a gate rather than a report.
@@ -20,30 +23,94 @@ indistinguishable from one that cannot.
| Check | Asserts | Found |
|---|---|---|
| `links` | every relative link resolves | — (run ad hoc during authoring; now permanent) |
| `rests-on` | `decisions:` and `extends:` name records that exist and are **accepted** | the class behind both incidents |
| `live-citation` | a governing document citing a **superseded** record names its replacement in the same paragraph | `01-to-be/README.md` citing ADR 0022 as live guidance |
| `supersession` | if A says it was superseded by B, B says it supersedes A | ADR 0012 never declared that it superseded 0011 |
| `numbering` | the number in the filename is the number in the heading | — |
| `topics` | every record names a topic the index knows | — |
| *(index.py)* | the written reading order matches what the records say | — |
| `no-records` | `02-DECISIONS/` and `01-RESEARCH/` hold their README and nothing else: a decision record or research effort lives in the knowledge base (ADR 0290 rule 9) | written with the change that removed the records (issue 426); it failed on a record file put back in `02-DECISIONS/` before it passed |
| `decisions-form` | every entry of a design's `decisions:` is "ADR NNNN", four digits | written with issue 426; it failed on an entry `ADR 290` before it passed |
| `mention-form` | every "ADR N" in the text has four digits, and every "research N" three; a code span is an example and is not read | written with issue 426; it failed on `ADR 12` and `research 45` before it passed |
| *(index.py)* | the reading order's six topics are written in `02-DECISIONS/README.md`, none twice, and the README stores no list of records (ADR 0248). The knowledge base reads these lines and refuses a record whose topic is not one of them | a stored list made every two open decisions conflict ([issue 297](../../04-ISSUES/297-two-open-decisions-always-conflict/00-report.md)); it failed on the README of `main` before the list was removed |
| `status-vs-code` | a to-be document naming specific code is not still `designed` | **ten documents**, several with a *What was built* section, describing lab-proven code |
## Where the rules on records went
The decision records and research efforts left this repository for the knowledge base (ADR 0290, issue 426),
and the rules these checks held over them left with them. The knowledge base holds each as a rule of its own,
named in its refusal: `knowledge-base.propose` refuses a record that breaks one of the rules that judge one
record, naming every reason at once, and the four rules that judge the whole record set run after every
acceptance and raise a condition when one fails (mesh-knowledge's `rules.go`, one table-test case per rule).
| Rule here until issue 426 | Where it lives now |
|---|---|
| `records.py``rests-on`, for a record's `extends:` | the knowledge base's `rests-on` at propose, and the whole-set rule `rests-on-live` |
| `records.py``rests-on`, for a design's `decisions:` naming an accepted record | **nowhere yet**: this repository cannot read the knowledge base. The reviewer checks each new entry with `knowledge-base.get`. Designs become the knowledge base's own kind under issue 428 |
| `records.py``live-citation`, in a record | the knowledge base's `live-citation` |
| `records.py``live-citation`, in `00-META/` and `03-DESIGN/01-to-be/` | **nowhere yet**, as above: the reviewer checks a cited record's status with `knowledge-base.get`; issue 428 |
| `records.py``supersession` | the knowledge base's `supersedes-accepted` and `self-link`, and the whole-set rule `supersession` |
| `records.py``partial-supersession` | the knowledge base's `changes` link to a clause, `changes-in-force` and `supersedes-accepted`, and the whole-set rule `partial` |
| `records.py``clauses` | the knowledge base's `has-clause`, `clause-numbering`, `clause-lead` and `clause-size` |
| `records.py``numbering` | the knowledge base's `title` and `reserved` |
| `records.py``topics`, `index.py` (a record's topic) | the knowledge base's `topic`, against the reading order `index.py` still checks |
| `records.py``insights` | the knowledge base's `insight-form`; an accepted record changes by a new version marked as a progressive insight |
| `index.py --print` | `knowledge-base.holds` with a topic |
| `cycle.py` research: a known status, `became:` when graduated | the knowledge base's `research-status`, `overview` and `became` |
| `cycle.py` no two decision records share a number | the knowledge base's `reserved`, and the issue tracker's counter behind `knowledge-base.reserve` |
| `cycle.py` an accepted decision cited by nothing | the knowledge base's whole-set rule `cited`: a record accepted there is cited by another record or names a home (`homes` at propose) |
| `cycle.py` a deferral names the issue that tracks it | the knowledge base's `deferral` |
| `words.py` over research efforts from 2026-10-07, and `kept` in `words-allowed.md` | the knowledge base's `retired-word` and `daemon-line`, from this glossary |
## What is deliberately not checked
- **`02-DECISIONS/` and `01-RESEARCH/` may cite superseded records freely.** A decision record
discusses history; research records what was observed. Flagging those would produce noise on
correct documents, and a check that cries wolf gets suppressed — which costs more than not
having it.
- **`03-DESIGN/00-as-is/` may rest on a superseded record.** It describes what runs, and what
runs was built under whatever was decided at the time
*as-is describing a superseded decision is exactly what as-is is for*).
- **Whether a citation's prose is still true.** Only whether the record it points at is live.
A document can cite an accepted record and describe it wrongly, and nothing here notices.
So "governing" means `00-META/` and `03-DESIGN/01-to-be/` — the documents that tell somebody
what to do.
- **Whether a record a document cites exists, or is still accepted.** A checkout cannot read the knowledge
base, and a check that called the mesh would make every merge here depend on it. Until designs become the
knowledge base's own kind (issue 428), the reviewer asks `knowledge-base.get` for each record a change
cites in `00-META/` or `03-DESIGN/01-to-be/` (the documents that tell somebody what to do). An as-is
document may rest on a superseded record: it describes what runs, and what runs was built under whatever
was decided at the time (ADR 0006: *as-is describing a superseded decision is exactly what as-is is for*).
- **Whether a citation's prose is still true.** A document can citean accepted record and describe it
wrongly, and nothing here notices.
## Adding a check
State what incident it would have caught, and make it fail before you make it pass. A check
whose failure has never been observed is a guess about its own correctness.
## cycle.py
The development cycle, checked (ADR 0080):
every design parses, its `layer:` matches its folder and its `status:` is known; a to-be design names a
decision in `decisions:`; an in-progress or implemented design names its owning code; and no two issue
records share a number (issue 155 — the number is how a record is cited, and `main` lags every open pull
request, so two people reading it allocate the same one). An issue report carries the day it was `opened:`
and no status key — `status:`, `located-in:`, `fixed-by:`, `amended-design:`, `replay:`, `replay-none:` or
`resolved:` — because an issue's status lives in the issue tracker (ADR 0275); it failed on all 361 reports
that carried them before they were removed. The rules on an issue's status this check held until then — an
owner for a located issue, a fix for a resolved one, and a core issue's `replay` or `replay-none` (ADR 0237)
— are refused at the write by `issue-tracker.move`. Its rules on research efforts and decision records left
with them (the table above). `python3 00-META/checks/cycle.py`
## words.py
One word per thing, checked (ADR 0244).
It reads the glossary's *Not:* lines (the retired words, each with its scope) and *Identifier until
renamed* lines, and fails on a retired word or a bare identifier in running prose — what is left once code,
quotations, struck-through text, link targets, comments and frontmatter are taken out — in `00-META/`,
`03-DESIGN/`, `AGENTS.md`, `README.md`, and issues opened from 2026-10-07. Decision records and research
efforts live in the knowledge base, which holds a proposed record to the same words (the table above). It
also fails when a glossary head word heads two entries or is also retired.
[`words-allowed.md`](words-allowed.md) names a document that could not be reworded at once, with a date. With `MESH_CATALOG_DIR` set it compares the catalogue's copy
of the tools' retired words (`retired-words`) with the glossary.
It also fails on a **daemon's line** (ADR 0272,
its insight of 2026-10-09): the operator types one command, `nox`, and a document names a `nox` line, never the
controller's or the node-engine's binary followed by one of its commands. The lines are the check's own list
(`DAEMON_LINES`); a command that is a line only with an argument after it is matched as a line is written. Looked
for in the same documents, from 2026-10-09 for issues, in the document's own words — code blocks and
spans included, quotations and block quotes taken out. With `MESH_CATALOG_DIR` set it compares the catalogue's copy
(`retired-lines`) with the list.
It failed on something real before it passed: 581 uses of retired words in 63 documents on its first run, besides the glossary itself and research 034 —
"the host" for the node-engine in 30 designs, "control plane" in 10, the glossary's own entry for the tool runner —
and two glossary contradictions (the "console", the deprecated broker's seat), fixed in the change that
added it. A retired word's plural (`s` or `es` on its last word) is matched as the word: until 2026-10-07
it was not, so "alerts", "rollouts" and "the hosts" (the node-engines) passed while their singulars
failed; the plural failed on 15 uses in 10 documents before it passed. A verb that reads like a plural
("a node hosts") is a finding too, and is reworded. Homonyms (*plan*, *gate*, *tier*, *ask*, *store*, *record*, *check*) are not checked by it: a
word list cannot tell one sense from another, so they are reviewed.
@@ -37,13 +37,13 @@ incident behind it is not written down, and the fix is to write it down, not to
| Rule | What it means |
|---|---|
| **Never write to a production database directly** | No insert, update, delete or schema statement executed against production by hand. Schema changes go through numbered migrations; data changes go through application code or the module's own capabilities. Raw statements skip every side effect the proper path has — events, audit, cache invalidation, fan-out. |
| **Every schema change is a migration** | Numbered, in the module's own language, compiled with it. Both a baseline for a fresh installation *and* an incremental migration for installations that already exist. If code references a column, the migration creating it must exist. [ADR 0013](../02-DECISIONS/0013-schema-changes-are-numbered-migrations.md) |
| **Every schema change is a migration** | Numbered, in the module's own language, compiled with it. Both a baseline for a fresh installation *and* an incremental migration for installations that already exist. If code references a column, the migration creating it must exist. ADR 0013 |
| **Never bypass the pipeline** | No manual database edit, no manual restart as a workaround. Fix the cause and deploy. A workaround that works is a workaround that is never removed, and the next person cannot tell the node from its declaration. |
| **Never create a symlink** | A hand-made link caused production data loss through container volume resolution, and the judgement needed to make a safe exception is exactly the judgement unavailable at the moment it matters. **The mesh creates none at all** ([ADR 0012](../02-DECISIONS/0012-the-mesh-creates-no-symlinks.md), which supersedes [ADR 0012](../02-DECISIONS/0012-the-mesh-creates-no-symlinks.md)). The links the installer still reconciles are a migration, not a permission. |
| **Never create a symlink** | A hand-made link caused production data loss through container volume resolution, and the judgement needed to make a safe exception is exactly the judgement unavailable at the moment it matters. **The mesh creates none at all** (ADR 0012, which supersedes ADR 0012). The links the installer still reconciles are a migration, not a permission. |
| **Never push directly to the main branch** | Branch, push, review, merge. Every merge is a human checkpoint, without exception — **including in this repository**. A documentation repository is not a lower tier of care; a decision record lands the same way a service does. |
| **One change per pull request, and never merge unapproved work** | Unrelated improvements bundled together cannot be reviewed or reverted separately. And the checkpoint is **a person deciding, not a person clicking** — work may be merged by whoever wrote it once a human has explicitly approved *that merge*, and never on a standing permission, an instruction to do the work, silence, or the author's own judgement that it is ready. [ADR 0023](../02-DECISIONS/0023-approval-is-the-checkpoint.md) |
| **One change per pull request, and never merge unapproved work** | Unrelated improvements bundled together cannot be reviewed or reverted separately. And the checkpoint is **a person deciding, not a person clicking** — work may be merged by whoever wrote it once a human has explicitly approved *that merge*, and never on a standing permission, an instruction to do the work, silence, or the author's own judgement that it is ready. ADR 0023 |
| **Never open a pull request unprompted** | A permissions list saying it is allowed is not a request. |
| **A failed step fails the job** | A sequence that continues past a failure does the next thing in the wrong place. Gate each step on the last. [ADR 0010](../02-DECISIONS/0010-delivery.md), and §5. |
| **A failed step fails the job** | A sequence that continues past a failure does the next thing in the wrong place. Gate each step on the last. ADR 0010, and §5. |
### A failed step must stop the steps after it — how it was earned
@@ -69,7 +69,7 @@ reported failure, nothing stopped, and the damage happened somewhere nobody was
- **Every runtime variable is declared.** A variable the module reads and the manifest does not
declare is invisible to the mesh: it will not be generated, injected, or audited.
- **Provisioned credentials arrive through declared requirements**, never hardcoded in code,
compose files or scripts. [ADR 0009](../02-DECISIONS/0009-modules-and-the-graph.md)
compose files or scripts. ADR 0009
- **Never install a package by hand.** A package is declared in the manifest and arrives the
way every other package does. A hand-installed package is invisible to the mesh: it is not
declared, not reproduced on the next node, and not present after a rebuild — and the node
@@ -82,7 +82,7 @@ reported failure, nothing stopped, and the damage happened somewhere nobody was
- **Every standalone application gets its own repository**, with a manifest at its root,
registered as a build source. Creating an application directory in the monorepo is a
@@ -33,7 +39,7 @@ how-we-build.md ──► constitution sync ──► knowledge base ──► i
| Layer | What it is | Changes when |
|---|---|---|
| `00-as-is/` | The mesh that exists today. Describes shipped behaviour, including behaviour nobody would choose again. | Something ships, or an as-is claim is found to be wrong. |
| `01-to-be/` | The mesh being built toward. Every statement traceable to a record in `02-DECISIONS/`. | A decision is taken or amended. |
| `01-to-be/` | The mesh being built toward. Every statement traceable to a decision record in the knowledge base. | A decision is taken or amended. |
A to-be document that ships does not move. Its as-is counterpart is written or updated, the
to-be document's frontmatter goes to `implemented`, and both stand — one describing what runs,
@@ -49,12 +55,32 @@ the expensive half.
| [03](03-issues.md) | Issues | Something is wrong — often with the owner unknown |
| [04](04-build-handoff.md) | Build handoff | A design is ready to be built |
| [05](05-constitution-sync.md) | Constitution sync | `how-we-build.md` changed a rule the mesh enforces |
| [06](06-writing-a-module.md) | Writing a module | Something that runs today must run on the mesh |
| [07](07-feature-branches.md) | Feature branches across repos | Work that changes code, in one repo or several at once |
## Status lives in frontmatter
## The cycle is checked
Research overviews, design docs, issue reports and decision records each carry their status as
YAML frontmatter (schemas in the section READMEs and playbooks). There are **no central status
The flow above is a rule, and a rule states how it is checked:
[`00-META/checks/cycle.py`](../checks/cycle.py) refuses a to-be design that names no
decision, an `in-progress`/`implemented` design that names no owning code, two issue folders
sharing a number or an issue report carrying a status key; a `graduated` research effort that
does not say what it became is refused by the knowledge base's `propose` (ADR 0290). Run it with `records.py`, `index.py` and
`words.py` before any HQ merge (`sh merge-check.sh` runs all four). The issue rules the check
once held — `located` with an owner, `resolved` with a fix, a core issue with its replay — are
refused at the write by the issue tracker's `move`
(ADR 0275). What the checks cannot see — that code work actually started from a
handoff — is held by playbooks [04](04-build-handoff.md) and [07](07-feature-branches.md):
a feature branch exists because a design or an issue sent it.
## Status lives in frontmatter, an issue's in the tracker
Design docs carry their status as YAML frontmatter (schema in playbook [02](02-graduation.md)).
Research efforts and decision records carry theirs in the knowledge base, which refuses a wrong one at
`propose` (ADR 0290). An issue's status lives in the
issue tracker, which refuses a wrong status at the write, and its report here carries only
the day it was opened (playbook [03](03-issues.md)). There are **no central status
files** and no decision ledger. **Every decision is a record in
[`02-DECISIONS`](../../02-DECISIONS/)** — if it is worth recording it is worth a record, and if
it is not worth a record it is not recorded. Cross-cutting views, the decision index included,
are generated on demand by the `hq-status` skill and never written to disk.
the knowledge base** — if it is worth recording it is worth a record, and if
it is not worth a record it is not recorded. Cross-cutting views are generated on demand by the
`hq-status` skill and never written to disk (ADR 0248); the decisions of a topic are
**Trigger.** Something that runs today must run on the mesh, or a new capability must be
declarable.
**Who runs it.** Whoever is porting or writing it.
*Written 2026-09-01 from doing this for the first time end to end. Every step below exists
because skipping it cost something.*
## Before anything: read what runs
**A module is written from the thing, not from memory of the thing.** For a port, that means its
current compose file, its environment, and where its data actually sits. Assumptions about any of
the three have been wrong every time they were not checked.
Three questions, answered from the machine:
| | why it decides something |
|---|---|
| **what containers, and how do they find each other?** | more than one means a `network`; names between them must match what the software is configured to dial |
| **where is its data?** | a bind mount moves with a path; a named volume does not; an anonymous volume is already losing data on every redeploy |
| **which values are secret, and which are merely settings?** | a secret goes in `own-secrets` or a grant; a setting goes in the manifest and may be overridden per node |
## The steps
1.**Name what it provides and requires**, if anything. A name is what a consumer is coupled to,
not the role it plays (ADR 0027):
`postgres-database`, not `database`. Most modules provide nothing and require nothing — an
application is usually a leaf.
2.**Declare capabilities, not dependencies, for facts about the machine.**`container-runtime`,
`package-manager`, `seat`. A capability is detected and refused against; it is not something a
module can install.
3.**Write the resources in the order they must happen.** They are applied in the order written
and orphans are removed in reverse, so a `network` is written before the containers that join
it and removed after them.
4.**Put every secret in a file, never in `env`.** A declaration travels over the broker in plain
text: a password in `env` is a password the broker sees. **The mesh delivers parts; a module
that needs them combined combines them.**
**A sealed file holds the password and nothing else** — no key, no `=`, no newline that means
anything. So `env-file` must never point at one. It points at a file the module *declares*,
whose content leaves a hole:
```
own-secrets superuser → /var/lib/postgres/superuser.secret the password, alone
a file /var/lib/postgres/superuser.env, mode 0600,
content: POSTGRES_PASSWORD=${secret:superuser}
the container env-file: [/var/lib/postgres/superuser.env]
```
The node-engine fills the hole on the machine, which is the only place both halves exist — the mesh
discarded the value
([credentials and their rotation](../../03-DESIGN/01-to-be/13-credentials-and-their-rotation.md)).
A **provisioner** is the exception: it reads a password file, so it mounts the `.secret`
directly.
Every example module in `mesh-controller` had this wrong and shipped: `own-secrets` pointing at a
path *named* `.env`, mounted as `env-file`, holding a bare password. Docker reads that as a
malformed line and the container starts **with no password set at all** — not a failure to
start, a service running on the wrong credential. They parsed and they resolved. Two tests in
`examples/modules` now refuse both halves of it.
Add `restart-on` naming the env file, or the container keeps the credential it started with
through every rotation.
5. **Pin every image by digest.** A tag moves. The manifest in a repository names artifacts; the
manifest the mesh holds names digests, and they are not the same document.
6. **Decide generate or accept.** A new module's credential is generated. **An adopted one keeps
the credential it already has** — `secret accept` — because minting a new password for a
database that already exists locks the application out of its own data.
7. **Add a provisioner only if the software cannot read a file.** A proxy that watches a
directory needs nothing. PostgreSQL needs `CREATE ROLE`, an object store needs a bucket and a
policy, an identity provider needs a realm and a client — those need a small program beside
them. It reads what the mesh granted and reconciles; it does not decide anything.
8.**Prove it in the lab, against the real software.** Not that a container started — that the
thing works: the credential authenticates, a wrong one is refused, the containers reach each
other, the data survives a restart.
## What the first port actually cost
Six attempts, one real bug. Recorded because the ratio is the lesson: **the mesh was right every
time and the scaffolding was not.**
- A shape existed in the language and no host implemented it, so every declaration carrying one
was refused whole — correctly, and the node-engine said exactly that. **Nobody was reading the node-engine's
log.** Read it first; it is the only place that says why a machine did nothing.
- A blind find-and-replace renamed a provision in quotes and missed the same word bare.
- A command was tested only for the invocations that should fail, so it rejected every real one
and the suite stayed green.
- A test asserted on a helper rather than on the code that calls it, three separate times. **A
test that cannot fail when the behaviour is deleted is not defending the behaviour.**
## Rules
- **Read the node-engine's log before theorising.** A declaration that was sent and not applied says so
there and nowhere else.
- **A failing test is kept, not skipped.** It is the reproduction.
- **Never rotate during an adoption.** Rotation is a separate act, afterwards, deliberately.
- **A data directory is never removed by the mesh** (ADR 0030),
and that protects against the mesh only — not against a disk or a mistaken command.
@@ -15,27 +15,30 @@ and a forge address is an operational detail (see [`README`](../README.md)).
| Repository | Owns |
|---|---|
| `hal` | The monorepo — the node runtime, the module catalogue, the delivery machinery, and the bootstrap scripts. Every core module lives here. |
| `hq` | This repository, under the company organisation — mission, research, design, decisions, issue diagnosis. Company-scoped ([ADR 0019](../02-DECISIONS/0019-how-this-repository-works.md)); the mesh is its first product. The source of truth for *why*. Carries no implementation. |
| `hq` | This repository, under the company organisation — mission, research, design, decisions, issue diagnosis. Company-scoped (ADR 0019); the mesh is its first product. The source of truth for *why*. Carries no implementation. |
| *(one per application)* | Every standalone application, site or side-project gets its own repository, with `module.yml` at the root. Registered with the mesh as a build source; built and deployed by the same pipeline as anything in the monorepo. |
| `migration` | **Private.** The record of one installation replacing the predecessor mesh with this one: the runbook, a dated log of every step and what it cost, the per-service data procedures, the readiness checks, and the scripts. Private because it is the opposite of this repository in every way that matters — it names machines, addresses, ports and paths, because a procedure that cannot be followed is not one. Where hq asks *what did we decide and why*, that repository answers *what happened on the machines, in what order, and what to do next*. Its `HANDOFF.md` is where somebody picking the work up starts. |
## What the mesh becomes
[ADR 0019](../02-DECISIONS/0019-how-this-repository-works.md) records the repositories the
monorepo decomposes into. **`mesh-lab`, `mesh-host` and `mesh-control` exist so far** — the lab is built first
([ADR 0016](../02-DECISIONS/0016-the-lab.md)); the rest are the
so far** — thelab was built first (ADR 0016). The tiered
decomposition below is the planned shape; the repositories built to date do not map onto it
one-for-one — `mesh-catalog`, `mesh-sdk` and `mesh-tools` exist where the table names
`mesh-foundation` and `mesh-surfaces`, and reconciling the two is itself still ahead.
| Repository | Tier | Holds |
|---|---|---|
| `mesh-host` | 0 | **exists.** The node host — one statically linked binary, requiring nothing present ([ADR 0005](../02-DECISIONS/0005-the-node-host.md)) |
| `mesh-substrate` | 1 | the four pinned services, as declarations |
| `mesh-control` | 2 | **exists.** The control plane and its contexts — one of seven built ([ADR 0006](../02-DECISIONS/0006-the-substrate-and-the-control-plane.md)) |
| `mesh-host` | 0 | **exists.** The node-engine — one statically linked binary, requiring nothing present (ADR 0005) |
| `mesh-foundation` | 1 | the four pinned services, as declarations |
| `mesh-controller` | 2 | **exists.** The controller and its contexts — one of seven built (ADR 0006) |
| `mesh-surfaces` | 3 | tools, web, cli |
| `mesh-sdk` | — | contracts shared across tiers |
| `mesh-sdk` | — | the stable spine modules build against — the tool-serving harness, the messaging/event framework, the contracts and core primitives. Holds nothing per-module and nothing volatile (ADR 0039). |
| `mesh-lab` | — | **exists.** The lab — scenario lifecycle, networking, placement. Ships to nobody; runs on a workstation. |
Tier 4's shape is open, and deliberately so: see ADR 0019 and
- **Areas touched:** every `hal/*` and `noxflow/*` module; the pipeline's dependency
graph; the knowledge base; agent identity and credentials.
## Summary
HAL has 124 modules. That number is not a maintenance problem in itself — it is the
**symptom of missing bounded contexts**. Modules are split not because they model
different domains, but because splitting is the only lever the platform offers:
- no way to run one daemon on one node without making it a module
(`hal/claude-licences` — one daemon, single-node)
- no way to expose two of a kind from one module
- no namespace separating the mesh from the software it runs
This effort establishes the **current state**, the **ideal state**, and the sequence
between them.
## Trigger
A night of debugging that produced four fixes and one conclusion. Every fault was a
boundary fault:
- Per-agent Claude credentials had to be written by the *noxflow runtime*, because
`agents.claude_account` is in noxflow's database — even though agent identity is a
mesh concept and node identity already lives in the mesh registry.
- Whether `hal/brain` may depend on noxflow took three attempts to answer, twice
wrongly, because the ownership boundary was never stated.
- Authoritative documentation existed in `mesh_docs` and was not found, while a
proposal in repo markdown was invisible to search entirely.
## Decisions taken (2026-08-22)
| Question | Decision |
|---|---|
| What should noxflow become? | Decompose into `hal/*` modules; noxflow returns to tasks/workflows |
| Who owns agent identity? | `hal/agents` — a mesh concept, alongside nodes |
| Where do third-party apps live? | Out of this repo. They run *on* the mesh; they are not *of* it |
| Knowledge structure | Modelled on `papa-hq`; implementation choice left open |
## Open questions
Tracked in [`analysis.md`](analysis.md) under "Open questions".
## Deliberately not decided
Recorded so they are not mistaken for oversights. Each is open, and each comes out of
[ADR 0001](../../02-DECISIONS/0001-mesh-brokers-nodes-host-agents-think.md); this effort stays
`active` until they are answered.
| Question | Status |
|---|---|
| `hal/scheduler` — infrastructure, or part of the work context. | Open. |
| Which context owns the executor. | Open. |
| Catalogue destination — one repository or many. | Open. Phase 4. |
| What the shared library keeps after extraction. | Open. Phase 3. |
| Where human agent modality is recorded — which user, on which node, a human agent acts as. | Open. Required by the model; not yet stored. |
| Which domains the modules outside the platform core group into. | Open, from [ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md), which settles the principle and deliberately not the list. |
Eight contexts instead of 33 platform modules, with the catalogue elsewhere.
The naming corrects itself in the process: what is called "brain" today is
infrastructure, and the word survives only as a concept an Agent owns — never again as a
module name.
This shape depends on **named features with per-node opt-in** — without it, every
independently-deployable unit inside a context becomes a module again and the count
returns. That mechanism is the subject of a separate ADR.
## Decided since
| question | decision |
|---|---|
| Meetings — own context or part of agents? | **Neither.** A meeting is a thread in the record; `hal/stream` absorbs it. The category was wrong. |
| Thoughts — autonomy or agent behaviour? | `hal/agents`. Any agent can have a thought. |
| `hal/axon` vs `hal/cortex` | `hal/cortex` is vestigial — no `daemon/` source, unit inactive on every node, description copy-pasted from axon. Messaging belongs to `hal/stream`. |
| `hal/hypothalamus` | Orphaned build output. A `dist/` directory with no `module.yml`, so not a module at all. |
| Who owns agent identity? | `hal/agents`. Nodes and agents are decoupled; a node is only a place an agent can run. |
| Where do third-party modules live? | Outside this repository. They run *on* the mesh, not *of* it. |
## Open questions
1.**`hal/scheduler`** — infrastructure (fire an event later, belongs to `hal/mesh`) or
part of `hal/work`?
2.**The executor.** Pulling and routing work items is `hal/work`; spawning a session is
`hal/agents`; providing the place to run is `hal/mesh`. It currently straddles all
three, which is the same kind of straddle that put credential-writing in the noxflow
runtime.
3.**Catalogue destination** — one repository, or per-application repositories, given the
pipeline resolves dependencies across the registry rather than the filesystem?
4.**SDK residue** — after extraction, does `hal/sdk` keep transport (`amqp-client`), or
does that belong to `hal/stream`? Everything imports it, which argues both ways.
5.**Human agent modality.** ADR 0001 requires a fact the mesh does not record: which
user, on which node, a human agent acts as. Where does it live — an attribute of the
| human-modality surfaces — brain **in cortex mode**, desktop notifications, desktop environment | on the host, by definition |
| module stacks | already Docker; systemd layer is the redundant part |
This split is not a compromise between the options. It is what the domain model implies,
and it is a decent sign that the model is doing work.
---
## 6. Options, with costs
| | option | cost | what it buys |
|---|---|---|---|
| **A** | **Keep systemd; systemd-in-container for Phase 0** | privileged containers, heavy images, slow iteration; local mesh keeps a shape production does not have | nothing changes in production; smallest change to the mesh |
| **B** | **Drop the `hal-module@` layer only** — let Docker's restart policies supervise stacks directly | reimplement uniform start/stop across ~12 call sites; lose rate-limited restart and pull-failure recovery | removes the redundant layer; does **not** solve Phase 0 on its own, since the daemons still need supervising |
| **C** | **Containerise the mesh daemons; Docker supervises** | container-aware `systemd_journal`/health; host access for daemons that write host files; the human-modality surfaces stay on the host regardless | terminates the fate-sharing recursion without writing a supervisor; makes local and production the same shape; Phase 0 becomes much less of a special case |
| **D** | **Write a mesh-native supervisor** | the fate-sharing recursion (§3), plus matching systemd's maturity — backoff, resource limits, clean SIGTERM (`noxflow/runtime` already depends on `TimeoutStopSec=60`) — on machines that are somebody's daily driver | most "mesh"; least justified by the evidence |
**D is the option the phrasing "more hal mesh approach" points at, and the evidence argues
against it.** Supervision is not a domain concern the mesh is better placed to solve than
the OS; the mesh's distinguishing feature is brokering capabilities, not restarting
processes. C gets the benefit D is reaching for — the mesh not depending on host-specific
init — without the recursion.
**C and B compose.** C is the one that pays for Phase 0.
Worth noting what is *not* in this table: the dozens of `systemctl` call sites that
configure the **host OS's own** units — NetworkManager, resolved, oomd, zram, docker.service,
fail2ban, sshd, zfs, across the vendor-firmware and power modules, `wireguard`, `sshd`, `zfs`
and others. Those are not HAL supervising itself; they are HAL configuring an Arch box. They
are out of scope for every option above and do not go away under any of them.
---
## 7. An incidental finding
Documentation describes an automatic node rescue: `hal-rescue.sh:20` states it is "triggered
automatically by `hal-health.timer` when hal-meshware is failed".
**It is not.**`hal-health.sh` contains no call to `hal-rescue.sh` (checked, zero matches),
and **no unit in the repository declares `OnFailure=`** (checked, zero matches). The only
real triggers are the manual `rescue_node` tool
(`modules/hal/sdk/src/tools/deploy-rescue.ts:77-92`) and running `install.d/rescue.sh` by
hand.
This is worth recording for two reasons. It weakens any argument that systemd-adjacent
self-healing is already wired — it is not. And it is another instance of the pattern this
whole refactor is about: **a documented mechanism that does not exist, believed because it
was written down.** `00-META/how-we-build.md` calls this out as a rule; here it is again,
found by grep.
---
## 8. Open question
**Which supervision model does the mesh adopt?** A, B, C, D or a combination.
**This no longer gates Phase 0.** When this was written, the local mesh was assumed to be
built from application containers, which forced the question — there is no natural way to
run an init system inside one. The decision of 2026-08-22 to build development nodes as
**system containers** (see [`03-DESIGN/01-end-to-end-testing.md`](../../03-DESIGN/01-to-be/01-end-to-end-testing.md))
removes that pressure entirely: a system container runs a real init, so the existing model
works unmodified and the lab needs no answer here to exist.
What remains is the question on its own merits, which is worth keeping open because the
evidence above still holds: option B removes a layer that 44 of 44 module stacks have already
made redundant, and option C would let the mesh stop depending on host-specific init. Neither
is urgent. Both are now cheap to *try*, because there is somewhere to try them.
Option D — a mesh-written supervisor — remains the one the evidence argues against, for the
fate-sharing reason in §3.
## References
- [`002-local-mesh`](../002-local-mesh/analysis.md) — the effort this came out of
- [`02-DECISIONS/0001`](../../02-DECISIONS/0001-mesh-brokers-nodes-host-agents-think.md) — agent modality, which
decides what cannot leave the host
-`modules/hal/meshware/daemon/src/cerebellum.ts:815-828` — the self-restart workaround
-`modules/hal/meshware/systemd/hal-module@.service` — the per-module Docker lifecycle
reframes it: things that change together share an **authority**, not a package. This effort's own
measurement is what that record rests on — reachability being the *only* place modules genuinely
co-change is why connectivity is a context and why nothing else needed one.
**What to do with the ~50 modules that co-change with nothing** — *nothing.* They are modules.
Grouping is a tag and a query over the graph, neither of which anybody keeps true by hand.
**What remains is a sequencing question, not a grouping one**, and it moves rather than closes:
*whether applications leave the monorepo before or after they group* is
[research 009](../009-migration/00-overview.md)'s, because it is about how to get from here to
there rather than about what the shape is.
## Open (superseded by the above) questions
| Question | Why it is open |
|---|---|
| Whether provider modules group at all, and if so what a consumer's requirement names instead of a module. | The provisioning reference is load-bearing; getting it wrong is expensive. Opinion and evidence in the analysis; not yet decided. |
| Whether applications group into domains now and leave the monorepo later as a unit, or leave first. | Decided in principle — group first, then split — but the migration order has real cost either way. |
| What to do with the ~50 modules that co-change with nothing. | The evidence gives no grouping signal for them at all. That may mean they are correctly sized already. |
| Whether "group or leave" is even the right pair of options. | [Research 006](../006-mesh-from-scratch/code-skeleton.md) finds a third fate — **absorbed into the node host**, ceasing to be a module at all — and argues it is the correct answer for the reachability cluster this effort measured. If so, the cluster this effort found is evidence for absorption rather than for grouping. |
([ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md)). A
correction inherits the frame of the thing it corrects, and two of the mesh's oldest problems
look unsolvable from inside that frame:
- **The bootstrap circularity.** The mesh needs a database, a bus, a registry and an identity
provider. Those are modules the mesh installs. The mesh cannot install them before it exists.
This has been worked around repeatedly and never designed away.
- **Participation requires privilege.** Everything assumes root on a machine whose packages and
services the mesh owns. A phone cannot participate on those terms, and neither can a machine
someone else administers.
Designing from nothing is a way to find out which parts of the current shape are requirements
and which are residue.
**And there is more residue than expected, from a knowable source.** This began as a
**dotfiles repository** — the first two days of history adopt dotfiles, add per-node dotfile
overrides, and introduce service symlinking with an ignore file. The flat one-directory-per-tool
catalogue, linking rather than copying, adoption of already-configured machines, per-node
overrides, and the desktop modules are all inherited from that, not chosen for a mesh. Recorded
in [`03-DESIGN/00-as-is/10-module-catalogue.md`](../../03-DESIGN/00-as-is/10-module-catalogue.md).
That makes this effort's question sharper than "what would we do differently": much of what
looks like design is a generalisation of *place files on my machines*, never revisited because
it was never stated as an assumption.
## The requirements this is designed against
Stated by the operator, recorded here so the skeleton can be checked against them rather than
against taste:
1. The mesh manages multiple computers — **full control**, through modules installed to nodes.
2. Mesh state lives in a **database**: which modules on which nodes, logs, configuration.
3. Configuration has **several touchpoints** — tool surface, web interface, others — all hosted
by the mesh itself.
4.**Connectivity** is core: every node reachable from every other over a shared overlay, some
nodes publicly exposed, firewalls configured.
5. The mesh **hosts applications** — and requires some of them itself. This is the circularity.
6.**Arch Linux only for now**; ideally any device, including phones, on lighter terms.
7. The end goal is to **operate an IT company** on it — development, design, deployment, full
circle, self-hosted. Personal cloud infrastructure.
8.**Agents make it self-improving and self-healing.**
9. It is **end-to-end testable on one machine**
([ADR 0016](../../02-DECISIONS/0016-the-lab.md)).
## Status
A first skeleton exists, with four design moves that the current shape does not have. It is
`active` because two of them are unproven and one contradicts a record that is already
accepted.
**Finding worth stating up front:** [ADR 0001](../../02-DECISIONS/0001-mesh-brokers-nodes-host-agents-think.md)
names nine bounded contexts and **none of them owns connectivity** — no overlay, no resolution,
no firewall, no ingress. Requirement 4 has no home in the accepted decomposition, while
[research 005](../005-domain-grouping/analysis.md) found reachability to be the *only* part of
the catalogue where modules genuinely change together under one intent. The skeleton adds it.
## Open questions
| Question | Why it is open |
|---|---|
| Does the record — the event log contexts integrate through — belong to the substrate or the control plane? | It is infrastructure by shape and domain by content. Placing it wrong reintroduces a circularity. |
| One repository per tier, or per context? | Already open from ADR 0001 as "catalogue destination — one repository or many". The skeleton assumes per tier and does not settle it. |
| ~~Does an unprivileged node earn a place in the inventory, or only a presence?~~ | **Answered 2026-08-25** by the operator: a node is a *managed machine inside the mesh*, not an unprivileged something — and a disconnected node is still a node, in a different situation. The question posed a class distinction; the answer is that there is none, and what varies is **state**. Recorded as [ADR 0004](../../02-DECISIONS/0004-a-node-and-how-it-joins.md). |
| ~~Does absorbing overlay, filtering, packages, supervision and the container runtime make the host too large?~~ | **Answered 2026-08-25** — [`host-size.md`](host-size.md). Measured: the absorption is smaller than the machinery that already applies state, and eight of ten adapters already carry no dependency. The risk is not size but direction, and it is two modules wide. The claim survives with its scope corrected — the host carries one concern, *apply declared state on this machine*, of which the six are instances. Recorded as [ADR 0005](../../02-DECISIONS/0005-the-node-host.md), designed in [`05-the-node-host.md`](../../03-DESIGN/01-to-be/05-the-node-host.md). |
| ~~Four substrate services or five?~~ | **Answered conditionally**, which is the honest form — [`07-the-substrate.md`](../../03-DESIGN/01-to-be/07-the-substrate.md). The substrate is *what the control plane consumes and cannot grant itself*. The identity provider qualifies only if the control plane delegates authentication; if it authenticates natively it is an ordinary hosted service. The count follows from a decision not yet taken, and asserting four was asserting that decision. |
| Does `feature` survive? | The skeleton splits it in two and argues the conflation is what makes the delivery pipeline hard to reason about. Unproven. |
**Today's mesh assumes the second state from the first moment.** Its source, its packages and
its images are all expected to be self-hosted before there is anything to host them — which is
why raising a first node needs a script that exists solely to paper over the impossibility, and
why that script is the least-exercised path in the system.
Making the transition explicit has a second benefit: it is reversible. A mesh whose forge is
broken can re-bind delivery to external providers and keep improving itself while it repairs
the forge. Today that escape hatch does not exist, because the dependency is not expressed
anywhere it could be changed.
### So, concretely
One `gitea` module. One image-registry module. One package-registry module. Each a tier-4
workload. The mesh's own instances are distinguished from any other instance **by what they are
bound to, not by being different modules** — precisely the same answer as postgres, arrived at
by the same test.
## The five fates of a module
Research 005 asked whether a catalogue module should be **grouped into a domain** or **leave
the repository**. Working the tier test across the catalogue surfaces a third answer that
neither option covers, and it is the most common one.
| Fate | Means | Examples from today |
|---|---|---|
| **Absorbed into the host** | It is not a module at all. It is part of what "managing a machine" means, and belongs in tier 0. | overlay membership, packet filtering, package management, service supervision, container runtime, filesystem management |
| **Substrate** | The control plane cannot exist without it. Pinned, host-applied. | relational store, bus, object store, image registry |
| **Control-plane context** | It decides something across nodes. | connectivity policy, inventory, delivery, provisioning, observability |
| **Workload module** | The mesh hosts it. Grouped per [ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md). | media library, desktop session, collaboration tooling |
| **Leaves the repository** | A standalone application, per [ADR 0015](../../02-DECISIONS/0015-applications-live-in-their-own-repository.md). | the applications identified in research 005 |
**The first fate is the finding.** Research 005 measured the reachability cluster — proxy,
resolver, firewall, overlay — as the only place in the catalogue where modules genuinely change
together under one intent. The skeleton explains *why*: they are not four modules that ought to
be one domain module. They are four facets of one thing the host should own, currently
expressed as modules because a module was the only unit available.
Under this skeleton the overlay module and the firewall module **stop existing**. The host holds
membership and applies filtering; tier 2 decides the policy; the swappable backends stay
modules. That is a different and better answer than grouping them, and it was not visible from
inside the current frame.
It also partly answers research 005's other open question — the fifty modules that co-change
with nothing. Several are host concerns rather than domains: package management, container
runtime, filesystem tooling. Silence was the right signal after all; the wrong conclusion was
that grouping was the only available fix.
## What a module looks like on disk
Per [`skeleton.md`](skeleton.md) Move 4, a module declares **parts** — independently selectable
pieces of desired state — and produces **artifacts** — things built once per version.
```
<module>/
module.yml identity, what it provides, what it requires,
<domain>/ grouped per ADR 0022, list per research 005
mesh-lab/ the whole mesh, disposable, on one machine
mesh-sdk/ contracts shared across tiers — types, not behaviour
hq/ company-scoped, not a mesh repository — ADR 0019
```
## The dependency rule
**A tier may depend only on tiers below it.** Substrate never references the control plane.
The control plane never reaches into a node except through the host. A surface holds no logic
a second surface would have to reimplement.
This is the whole of the bootstrap answer, and per this repository's own rule it must say how
it is checked: a dependency-direction lint in the build, failing on an upward import. A tier
rule enforced by intention is the same as no tier rule — that is
[ADR 0010](../../02-DECISIONS/0010-delivery.md) applied to architecture.
## Move 1 — the substrate is applied, not delivered
**The problem.** The mesh needs a database, a bus, an object store and an image registry.
Today those are modules, and modules are installed by the delivery pipeline, which
needs the database and the bus. The first node is therefore raised by a special script that
exists only because of the circularity, and every later change to the substrate has to pretend
the circularity is not there.
**The move.** The host can apply a declaration without anyone telling it to. The substrate is
a **pinned bundle** the host carries: a fixed, versioned, self-contained descriptor of the
four services and nothing else. Raising a first node is `host apply substrate.lock` — not a
special path, just the ordinary one with no control plane on the other end.
The circularity disappears rather than being worked around: **the substrate is applied by tier
0, the mesh is delivered by tier 2, and they are different mechanisms on purpose.**
The price is real and should be named: the substrate is upgraded by bumping a pin and
re-applying, not by the pipeline. It gets less machinery than everything else — no per-node
selection, no provisioning, no fan-out — and that is the point. Five services justify a
simpler mechanism than a hundred.
## Move 2 — the host is one binary with capability profiles
**The problem.** Everything assumes root on a machine whose packages, services and network the
mesh owns. A phone cannot offer that, and neither can a work laptop. The current answer would
be a lightweight fork, which means two implementations and one of them rotting.
**The move.** One host binary, and a **profile** it detects rather than is told:
| Profile | Can | Typical |
|---|---|---|
| `managed` | packages, services, network, filesystem — the full surface | a machine the mesh owns |
| `user` | user-level services and tools; no package or network management | a shared or administered machine |
| `edge` | report presence, relay, expose a tool surface; hold nothing | a phone |
A module declares which profiles it can land on. Assignment to an incapable node fails at
declaration time, not at deploy time — a phone is not a machine that fails to install a
firewall; it is a node the firewall module cannot be assigned to.
This makes the phone case a **capability question rather than a platform question**, which is
what keeps it from becoming a second implementation. It also removes the current unstated
assumption that every node is equivalent — already false, and today handled by remembering.
## Move 3 — connectivity becomes a context
[ADR 0001](../../02-DECISIONS/0001-mesh-brokers-nodes-host-agents-think.md) names nine contexts
and none of them owns the overlay, the resolver, the firewall or the ingress. `config` owns
PKI, which is the closest thing, and it is not close.
Meanwhile [research 005](../005-domain-grouping/analysis.md) measured the whole catalogue and
found that reachability is the **only** place where modules genuinely change together under one
intent — the proxy with the resolver, the firewall with the overlay, repeatedly, because *how a
node is reachable* is one question asked in four places.
So the evidence and the gap point the same way. `connectivity` owns:
- the overlay every node joins, and the addresses on it
- name resolution, internal and public
- exposure — which services answer from outside, on which names
- filtering — what may reach a node at all
- certificates for both name spaces
This is an addition to an accepted record, so it is a decision, not a drafting choice. It
belongs in a new record that extends ADR 0001 the way
[ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md) does —
not written here.
### Where the networking actually lives
A context is an authority, not a running thing, so naming one does not say who brings the
overlay up. That splits three ways, and the split is the design:
| Concern | Tier | Why there |
|---|---|---|
| **Overlay membership** — this node joins, holds an address, keeps the tunnel up | **0, in the host** | Everything cross-node needs it *before* the substrate is reachable from elsewhere. A module cannot provide it, because installing a module is itself a cross-node operation. |
| **Policy** — who holds which address, what resolves, what is exposed, what is filtered | **2, `connectivity`** | Bookkeeping and authority. It decides; it runs nothing. |
| **Machinery** — resolver, reverse proxy, firewall backend, certificate issuance | **4, modules** | Swappable, and not every node needs them. A node without a reverse proxy is still a node. |
There is a second circularity hiding here, and it has to be closed explicitly: **the host's link
to the control plane does not run over the overlay.** If it did, the overlay would have to be up
before the host could be told how to join it. The link is ordinary outbound internet to a public
endpoint; the overlay carries node-to-node traffic only.
Joining is therefore: host lands → links out with a join token → control plane returns an
address and keys → host raises membership → the substrate on other nodes becomes reachable.
This also makes the `edge` profile honest rather than special-cased. A phone can hold overlay
membership in userspace without privilege, and cannot run the machinery. That is the profile
distinction doing its job.
## Move 4 — `feature` splits in two
The invitation was to check whether the concept survives. It does not, in one piece.
Today a **feature** means both *a thing built once* and *a thing selected per node*, and the
delivery pipeline is hard to reason about precisely because those have different cardinality
and one word ([ADR 0010](../../02-DECISIONS/0010-delivery.md)
is the pipeline half of the same confusion).
Split it:
| Concept | Is | Cardinality |
|---|---|---|
| **artifact** | something built and published — an image, a bundle, a package | once per module version |
| **part** | an independently selectable piece of a module's desired state | chosen per node, per assignment |
A module declares desired state in parts, and produces artifacts. An assignment names the node
and the parts. Delivery builds artifacts once and applies parts per node — and the two words
now carry the two cardinalities that the pipeline already has.
This keeps what features are genuinely for — per-node opt-in of *some* of a module, which the
work breakdown already calls for — and drops the conflation that makes the current model
confusing.
## What the mesh is, versus what it hosts
Tiers 0–3 are the mesh. Tier 4 is everything it carries, and the boundary is stated by
requirement rather than by taste: **a module is part of the mesh if removing it stops the mesh
managing nodes.** A media server does not. A relational store does — which is why it sits in the
substrate and not the catalogue, despite being, in every other respect, an application like any
other. An identity provider, notably, does **not**: the control plane authenticates its own
callers, so identity is a hosted service like the media server.
That test also settles the IT-company goal without a special category. Development, design and
deployment tooling are **workloads** — tier 4, hosted, provisioned, delivered like anything
else. The mesh does not grow a "company" feature; it hosts the tools a company runs on, and
the fact that it runs its own development on them is dogfooding, not architecture.
## What agents are, structurally
Self-improvement and self-healing are not a tier. Agents are participants
([ADR 0003](../../02-DECISIONS/0003-agents-are-persistent-employees.md)) that hold identity in
tier 2, act through tier 3 like any other caller, and run as workloads in tier 4.
This matters for one reason: **an agent must not have a privileged path**. Anything an agent
can do to the mesh, a person can do through the same surface, and anything it cannot express
through the tool surface is a gap in the surface rather than a reason for a back door. Self-
healing built on a private channel is unreviewable, and would be the one part of the mesh with
no human checkpoint.
## How this is tested
The lab ([ADR 0016](../../02-DECISIONS/0016-the-lab.md)) raises the
tree above on one machine: virtual machines as nodes, a real overlay between them, the real
substrate bundle, the real control plane, the real delivery path.
The tier rule is what makes that affordable. A scenario needing only tiers 0 and 1 is one
virtual machine and a pinned bundle — which is also, exactly, the bootstrap path. **The
hardest thing to test becomes the cheapest scenario to run**, and the first-node path stops
being the one thing nobody exercises until it breaks.
## What this skeleton does not answer
- Where the record lives. It is infrastructure by shape and domain by content, and putting it
in the substrate risks recreating a circularity in the one place the design just removed one.
- Whether tier 2's contexts are one repository or several. Open from ADR 0001 already.
- Whether an `edge` node is in the inventory or merely present — which decides whether "node"
is one concept or two.
- The migration. Nothing here says how today's mesh becomes this, and the skeleton is worth
little until that is costed.
## A naming near-miss, recorded
The tier-0 binary was first called `mesh-agent`, because "node agent" is the reflex everywhere
else in the industry. That is wrong here, and wrong in the specific way
[`how-we-build.md`](../../00-META/how-we-build.md) §4 exists to catch: **Agent** is a
first-class concept in this mesh — a participant, some of whom are human, holding identity and
memory ([ADR 0003](../../02-DECISIONS/0003-agents-are-persistent-employees.md)). One document
carried both meanings.
It is the same failure as the anatomy naming in the current runtime: an evocative domain word
pointing at infrastructure.
[ADR 0001](../../02-DECISIONS/0001-mesh-brokers-nodes-host-agents-think.md) supplies the fix in
its own title — *the mesh brokers capabilities; nodes host; agents think.* Three verbs, three
components: the control plane **brokers** (`mesh-control`), the tier-0 binary **hosts**
(`mesh-host`), the participant **thinks** (`agents`, untouched).
`mesh-node` was the alternative and was rejected: *Node* is the aggregate in the inventory — the
record of a machine — while the binary is what runs on it and does the hosting.
Provisioning is the mechanism the whole mesh rests on: a module declares what it needs, and the
mesh makes it exist, generates the credential, records the grant, and puts the values where the
module will read them. [ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md)
calls it the mesh's core concern rather than its plumbing.
[Research 006](../006-mesh-from-scratch/code-skeleton.md) then asks it to carry **more**: the
control plane becomes a consumer with its own requirements — a source of record, an image
registry, a package registry — satisfied by the same mechanism. That generalisation is only
safe if the mechanism is sound, and the as-is record says it is not, in named ways.
## Why now
Four weaknesses are already documented in
[`03-DESIGN/00-as-is/03-provisioning.md`](../../03-DESIGN/00-as-is/03-provisioning.md), each
observed rather than theorised:
- **Rotation has no fan-out.** A shared credential can be rotated without telling the peers
holding the old one. This has locked the mesh out of its own broker.
- **A grant is not a check.** The record says a resource was provisioned. Nothing verifies it
still exists, still has that credential, or is reachable from where the consumer runs.
- **A frozen password outlives its generation.** A generated secret written once diverges from a
persistent data directory initialised earlier, and presents as an authentication error.
- **No requirements is indistinguishable from provisioning that did not run.** A module that
declares nothing skips the stage, which is correct, and looks identical to failure.
Generalising a mechanism with these properties to the control plane's own dependencies would
make each of them fatal rather than annoying.
## The questions
| Question | Why it matters |
|---|---|
| What does a **grant** mean, exactly — a record that a resource was created, or a claim about the world that is continuously reconciled? | The difference between the current model and one where "provisioned" is checkable. Almost every weakness above is a symptom of the first answer. |
| How is a credential **rotated** with fan-out to every holder? | The mechanism grants easily and regrants not at all. This is the most damaging gap and it has taken the mesh down. |
| Can the **control plane** hold requirements, and what satisfies them before anything is installed? | The generalisation research 006 needs. Ties directly to the self-hosting transition. |
| What happens when a requirement **cannot** be satisfied — no provider, provider on an unreachable node, provider not yet installed? | Today this is silent or a stall. It should be a stated, visible state. |
| Does a requirement belong to a **module** or to one of its **parts**? | Research 006 splits `feature` into artifact and part. A part-scoped requirement means a database is not provisioned where the part that needs it is not installed. |
move one level up: the control plane holds what source exists and what has been built, and builds
the difference. **An event makes it fast; nothing makes it necessary.** The failures this effort
catalogued — a truncated commit list, a broken path match — become latency rather than silence.
**What is a deployed state?** — *Two comparisons, not an event.* Does every node's reported state
match what is declared, and is what is declared built from current source? A milestone can be
claimed by something that did not check; a comparison cannot.
**What produces a verdict, and what is it about?** — *An artifact, and it gates eligibility.* The
mesh must not converge onto something broken, so an artifact may be declared only once the lab
has judged it fit. Sharper than the question expected: a verdict is a property an artifact has,
not a report about a run.
**How does delivery work before self-hosting?** — *It mostly stops being a question.* A reconciler
reads source and writes artifacts; where those live is a binding, external first and internal
later. The transition looked hard because a pipeline's stages name their targets.
## What this effort was right about
Its first question — *what is a deployed state, and how does the mesh know it is in one* — was
marked "everything follows from this", and everything did. Both records above are answers to it:
0058 makes the applier the reporter, and 0063 makes currency a comparison. The effort put the
load-bearing question first.
## What is NOT closed by this
[ADR 0010](../../02-DECISIONS/0010-delivery.md) names four costs
and one of them is a real risk rather than a trade: **a reconciler that cannot reach its target
retries forever, and without something that notices, the failure is silence** — which is the
fault this effort exists to catalogue, reintroduced in a new place. That belongs to observability
and it is not designed.
## The questions (all answered above)
| Question | Why it matters |
|---|---|
| What is a **deployed state**, and how does the mesh know it is in one? | Everything follows from this. If a stage reports transport, "deployed" is a claim nobody checked. A desired-state model with reconciliation gives a different answer from a job-completion model. |
| Does the coordinator dispatch **stages**, or converge nodes on a **declaration**? | The current model is a state machine over stages. The alternative is that a node is told what should be true and reports what is. The second makes drift visible; the first cannot see it. |
| How does a change **become** a pipeline, reliably? | Detection has failed for reasons unrelated to the change, silently. |
| What produces a **verdict**, and what is it a verdict about? | Ties to the lab ([ADR 0016](../../02-DECISIONS/0016-the-lab.md)) and to a module carrying its own assertions. |
| How does delivery work **before self-hosting**, and across the transition? | From research 006: source and artifacts start external and are re-bound to internal providers. The coordinator has to be indifferent to which. |
| Does the **three-silo** split survive the artifact/part split? | [ADR 0010](../../02-DECISIONS/0010-delivery.md) is cardinality-driven, and research 006 renames the thing the cardinality is about. |
The migration appears to need exactly the capability the design declared it would not have.
The way out, to be tested: the new mesh does not adopt anything. It **declares** the workloads
from scratch and points them at data directories that already exist. Nothing inspects a running
machine to learn what is there; the declarations are written from the as-is layer, which is what
that layer is for. Data survives because it was never touched, not because it was adopted.
If that holds, adoption stays out of scope and the migration is ordinary declaration. If it does
not, adoption needs a one-time, explicitly unsupported tool, and that should be a decision rather
than a discovery.
## Sequencing
| Phase | What | Done when |
|---|---|---|
| **0** | **Build the lab's bootstrap scenario** ([ADR 0016](../../02-DECISIONS/0016-the-lab.md)) — virtual machines, a network, a way to place a binary, snapshot and reset. No forge, no coordinator, no pipeline. | A machine can be raised from nothing, reset, and raised again, repeatably. |
| A | Build tier 0, **inside the lab**. The host's interface first — it carries the skeleton's biggest unproven claim. | A bare machine becomes a managed node with no mesh present. |
| B | Build tier 1 and 2. The bootstrap scenario grows into the full one by addition — the same machines, with more placed inside them. | The lab raises a full mesh from nothing, repeatedly, from pinned external artifacts. |
| C | Enough of tier 3 to operate it. | The mesh can be driven without direct database access. |
| D | Declare the existing workloads against the new model. | The lab runs them, with copies of real data shapes. |
| E | **Rehearse the cutover in the lab** against a mesh built to resemble the real one. | Repeatable, and repeatably reversible. |
| F | Cut over. | The real nodes are managed by the new control plane. |
| G | Reach self-hosting — re-bind delivery from external providers to the mesh's own forge and registries. | The mesh builds and deploys itself. |
Phase G is deliberately last. Per research 006, self-hosting is a state the mesh **reaches**;
attempting the cutover and the self-hosting transition in the same move recreates exactly the
circularity the skeleton removes — and would mean a failed cutover could take away the means to
fix it.
## The questions
| Question | Why it matters |
|---|---|
| What state must **survive** the cutover, versus be re-created? | Workload data must. Provisioned credentials could be re-issued. The mesh's own inventory could be re-declared. Each answer changes the risk. |
| What is the **way back**? | A cutover with no rollback is not a plan. If workload data is untouched, reverting may be as small as re-pointing the old control plane at it — to be verified, not assumed. |
| How is the cutover **rehearsed** against something resembling the real mesh, without copying the real mesh into a repository? | The lab must be able to model the real topology's shape without carrying its identity. |
| Does anything have to keep running **during** the cutover? | Mail and source are the obvious candidates. If yes, "big-bang" is really "big-bang with exceptions", and the exceptions should be named now. |
| Is the forge inside or outside the cutover? | If the mesh's own forge goes down with the old control plane, the means of deploying a fix goes with it. This is the self-hosting circularity appearing as a migration risk. |
[The lifecycle design](../../03-DESIGN/01-to-be/03-scenario-lifecycle.md) closes on an open
question that is measurable rather than arguable:
> **What a snapshot costs.** Whole-scenario snapshots of several machines are the operation the
> inner loop repeats most, so their cost sets the loop's speed. If restoring is slow, the loop
> is slow, and everything above is theory.
Measured on a workstation, 2026-08-24. Numbers in [`measurements.md`](measurements.md).
## Why it matters
[ADR 0016](../../02-DECISIONS/0016-the-lab.md) makes the
bootstrap scenario the inner development loop for tiers 0 and 1 — the argument being that
raising a node from nothing stops being the least-exercised path and becomes the most-exercised
one. **That argument is only true if raising and resetting are cheap.** A loop that costs
minutes is a loop people avoid, and the least-exercised path stays least-exercised.
## Status
Measured, and the finding is a blocker rather than a data point.
**A snapshot on the current host is a full copy of the machine's disk.** 1.6 GB and ten seconds
for one small virtual machine at best — and over two minutes when observed a second time. Cost
scales with the number of machines and the size of their disks, not with what changed.
**With copy-on-write it is 0.13 seconds and costs the delta.** Verified, not assumed. The
projected four-machine reset cycle falls from roughly ninety seconds to roughly fifteen, of
which almost all is a boot that cannot be avoided.
The cause is not virtual machines and not incus. It is that the host offers incus exactly one
storage driver, `dir`, which has no copy-on-write and therefore no cheap snapshot. The kernel
supports btrfs; the userspace tool that would let incus use it is simply not installed.
So the question *"is the lab's inner loop fast enough"* currently answers itself the wrong way,
for a reason that is one declared package away from being fixed — and declaring packages is
something the mesh already does.
## Open questions
| Question | Why it matters |
|---|---|
| ~~How much does a copy-on-write pool actually improve it?~~**Measured: snapshot 9.9 s → 0.13 s, restore 10.4 s → 0.80 s, three snapshots sharing 1.36 GB rather than costing 4.8 GB.** The projected four-machine cycle falls from ~90 s to ~15 s. | Answered. The inner-loop argument holds *with* copy-on-write and did not without it. |
| How does the lab install its own prerequisites on a clean machine? | The lab needs a virtualisation daemon, copy-on-write tooling and a pool before it can do anything — and it cannot depend on the mesh for them, since it is where the mesh is built. |
| Why was the second snapshot more than twelve times slower than the first? | If snapshot cost is unpredictable rather than merely high, that is worse — a loop with a variable multi-minute step is one nobody trusts. |
| Does a scenario snapshot need the machines stopped? | Stateless snapshots of a running virtual machine capture the disk but not memory. Whether a mesh restored that way is coherent is not established. |
| What is the cost at scenario scale — four machines rather than one? | Only single-machine numbers were taken. If the operation is serial, four machines is four times the wait. |
Taken 2026-08-24 on a workstation with hardware virtualisation available, an NVMe-backed ext4
root, and 300 GB free. One virtual machine, 1 GiB memory, 2 CPUs, from a cached distribution
image.
## The environment, before anything ran
| Fact | Value | Consequence |
|---|---|---|
| Hardware virtualisation | present | virtual machines run at native speed; the choice in [ADR 0016](../../02-DECISIONS/0016-the-lab.md) is not paying an emulation penalty |
| Storage drivers the daemon offers | **`dir` only** | no copy-on-write, therefore no cheap snapshot |
| Host filesystems | ext4 throughout | nothing copy-on-write to put a pool on |
| btrfs kernel module | **available** | the kernel can do it |
| `btrfs-progs` | **not installed** | which is the entire reason the driver is absent |
The last two rows are the finding. The daemon advertises only `dir` because the userspace tool
for anything better is missing — not because the host cannot do better.
## Raising a machine
| Step | Time |
|---|---|
| launch call returns | **3.4 s** |
| machine actually usable — a command executes on it | **14.3 s** |
The gap matters for the lifecycle design: `raise` returning is not the same as the scenario
being ready, so the verb has to wait for the second number, not report the first. Reporting
the first would be the mesh's own recurring failure — transport reported as effect.
## Snapshot and restore
| Operation | Time | Disk |
|---|---|---|
| snapshot, first | **9.9 s** | **+1.6 GB** |
| snapshot, second | **> 120 s — did not complete** | — |
| restore call returns | **10.4 s** | — |
| machine usable again | **20.1 s** total | — |
Instance on disk before snapshotting: 1.5 GB. Snapshot directory afterwards: 1.6 GB. **A `dir`
snapshot is a full copy** — the storage cost equals the instance, and nothing is shared.
Implied copy throughput on the first snapshot is roughly 160 MB/s, which is far below what the
underlying NVMe can do and is consistent with a real, durable copy rather than a metadata
operation.
**The second snapshot is the more troubling number.** It exceeded two minutes and was still
running when the observation was cut off; only the first snapshot exists. Whatever the cause —
page cache exhausted by the preceding restore, writeback contention — the practical
consequence is that snapshot cost here is **not merely high, it is unpredictable**.
## What this projects to
A four-machine scenario, taking the optimistic single-machine numbers and assuming the
operations are serial:
| | one machine | four machines |
|---|---|---|
| raise, to usable | 14 s | ~57 s |
| snapshot | 10 s, 1.6 GB | ~40 s, 6.4 GB |
| restore, to usable | 20 s | ~80 s |
A reset-and-rerun cycle is therefore **around a minute and a half at best**, and unbounded at
worst, before any of the mesh's own work begins.
## The judgement
**This is too slow for an inner loop**, and the reason is not the design.
[ADR 0016](../../02-DECISIONS/0016-the-lab.md) argues that
making the bootstrap path the inner development loop turns the least-exercised code in the
system into the most-exercised. That argument holds only while resetting is cheap. At a minute
and a half a cycle, with occasional multi-minute stalls, the loop is one a person works around
— and the path stays under-exercised for exactly the reason it always was.
Nothing about virtual machines causes this. Hardware virtualisation is present and the machines
boot in fourteen seconds. **The cost is entirely the storage driver**, and the driver is absent
because one userspace package is not installed on the host.
The mesh already has the mechanism for that: a module declares a package, and a hook makes it a
working capability — which is precisely what was just done for the virtualisation daemon
itself, and what [`04-ISSUES/007`](../../04-ISSUES/007-an-installed-package-is-not-a-capability/00-report.md)
is about.
## The fix, measured
The comparison was subsequently run. One package — `btrfs-progs`, no dependencies — installed by
hand, the daemon restarted so it re-detected drivers, a copy-on-write pool created on a loop
file, and the identical image launched onto it.
| Operation | `dir` | copy-on-write | |
|---|---|---|---|
| snapshot | 9.9 s, then **> 120 s** | **0.13 s** | ~76× faster, and *consistent* |
grouping modules by domain. [Research 005](../005-domain-grouping/analysis.md) measured that
proposal and found its evidence holds in exactly one place — reachability — which
[ADR 0005](../../02-DECISIONS/0005-the-node-host.md) has since absorbed
into the host. The measured case for domain grouping has therefore been consumed by a decision
taken for unrelated reasons, and what remains is fifty modules that co-change with nothing.
That leaves the original complaint unanswered: the catalogue records **what was installed**
rather than **what anything is for**, and nothing in the system can see a relationship between
two modules. Grouping asserts relationships. A graph records them.
## Why now
A proposal to split modules into *provisioning services* and *applications* was worked through
and abandoned in favour of one concept with facets, for a reason worth keeping:
- The split cannot be filed consistently. A git forge is consumed as a service *and* operated
through a web interface. An analytics service grants tracking identity *and* is a dashboard
somebody reads. An identity provider grants authentication *and* has an admin console.
- The operator's correction is the sharper form: **what runs on the machine is a supervised
container, not something a user started.** That is a fact about *how a thing runs*, not about
what kind of thing it is — so it is a facet, not a taxonomy.
Filing decisions that follow from nothing are the disease research 005 measured. A second
taxonomy would reproduce it.
So [ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md) survives, and the question
becomes what a module must be able to **declare**.
## The shape being investigated
Five declarations, of which two exist today.
| Declaration | Today | Notes |
|---|---|---|
| **requires a resource** — a database, a bucket | yes | provisioning, [ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md) |
| **provides a resource** | yes | as above |
| **requires another module** | **no** | the dependency edge — the graph's substance |
| **excludes another module** | **no** | installing A makes B unavailable |
| **requires a node capability** | **no** | a graphical session, a container runtime, an architecture |
And one structural idea on top: **an interface module carries no implementation.** Adapters
provide it. An assistant interface with several model-provider adapters; a terminal interface
with several terminal adapters. A dependent names the interface and never an implementation.
**Prior art to measure against, not invent past.** This is a package manager's model, and the
platform's own package manager already has all of it: `depends` is the dependency edge,
`conflicts` is exclusion, and `provides` is the interface — several packages provide one
virtual name, and a dependent names the virtual one. That the design arrived at the same shape
independently is evidence for it. It is also a warning: dependency resolution, version
constraints, conflict handling and rollback are a long-solved and easily-botched problem, and
the effort should establish what to **delegate** rather than reimplement.
## Capabilities, and what may be installed
The operator's formulation: *system specs are capabilities, and capabilities unlock installable
modules — you cannot install a graphical application on a node with no display server.*
The question that follows is whether the mesh may install a capability. The effort's working
position, to be tested:
- **Intrinsic capabilities** — hardware, architecture, network position — are facts about a
machine. They are detected, never installed, and a module requiring one it does not have is
not unresolved but **impossible** on that node.
- **Provided capabilities** — a display server, a container runtime — are not a separate kind
of thing at all. They are modules that provide a capability, and requiring one is an ordinary
edge the graph resolves by installing it.
If that holds, "may the mesh install a capability" is not a policy question. It is dependency
resolution, and the only genuinely new thing is detection.
Which is where [issue 007](../../04-ISSUES/007-an-installed-package-is-not-a-capability/00-report.md)
bears directly: *an installed package is not a capability*. A provided capability is not real
because a package is present — it is real when it is present, running and working, and the
difference is exactly the class of fault this repository keeps recording.
## What it touches beyond the catalogue
The operator's assessment is that the machinery around a module is wanted and its integration
is not: **scheduled tasks, hooks, migrations, configuration and environment settings are worth
keeping; seeds are not; and the current integration is wrong enough to need a major refactor.**
That is a claim to test rather than adopt. Research 005 already found supporting evidence from
a different direction — that the densest apparent coupling in the catalogue is manifest
boilerplate churn, cross-cutting changes to the machinery applied N times — which is what an
integration being wrong looks like from the outside.
## Open questions
Struck-through rows are answered, with where. The rest are live.
### Settled
| Question | Answer |
|---|---|
| ~~What does the graph **delete**?~~ | For the existing system: nothing, it is already there ([`analysis.md`](analysis.md)). For the design: the module/resource distinction, the interface as a kind of thing, capability checking as a separate mechanism, domain grouping, and — the first clear deletion — **grant kinds**, once a module may only be granted what it exclusively owns ([`worked-provider.md`](worked-provider.md)). |
| ~~Is an interface a module, or a name?~~ | A **name**, and only where providers are genuinely substitutable. The adapter is what creates one; without an adapter there is a **tag**, which describes and does not bind ([`proposal.md`](proposal.md)). |
| ~~Where do domain modules fit?~~ | They do not. There is core infrastructure — concrete modules named individually, not flavourable, nothing standing in front of them. |
| ~~What happens to domain grouping?~~ | Superseded. Folders assert relationships; edges record them. What grouping was for is a tag and a query. [ADR 0009](../../02-DECISIONS/0009-modules-and-the-graph.md) is `proposed` and should be superseded rather than narrowed. |
| ~~Is there one kind of edge?~~ | **No — two.***Presence*, where a thing must exist, and *instantiation*, where a provider makes something for a consumer and hands back credentials. Instantiation implies presence, not the reverse. |
| ~~When two modules provide one name, who chooses?~~ | Neither the consumer naming a node nor the consumer not caring. The consumer declares the **scope of its own need** — shared across its instances, or one each — the mesh binds, and the binding is written down and sticky. Where it is written follows the scope. |
| ~~Can several modules share one database?~~ | **No.** A module is granted only what it exclusively owns — no shared writes and no read role on another's store, because reading couples you to its layout just as firmly. |
| ~~What about a dashboard reading a dozen stores?~~ | **The rule is about contexts, not processes.** The mesh's own board reading the mesh's own store is the mesh showing its own data — not a boundary crossing. Everything inside a context reads its store freely; what is forbidden is a *different* context reading it. An earlier answer here was wrong. |
| ~~Can every registry consumer be served another way?~~ | **Largely dissolves.** Of eighteen direct consumers, the owner keeps its database, node appliers are already stopped by ADR 0005, and the bulk are **foreign tenants** — thirteen tables across three contexts — who need to move out rather than read differently. |
### Live
| Question | Why it is open |
|---|---|
| How many instances should a module have? | Not derivable and not global: one-per-mesh is right for the broker and wrong for a store a disconnected node needs. It is per-module and nothing in the schema says it. |
| Can two of something coexist? | `excludes` covers part of it. Two terminals are fine, two things wanting one port are not, two brokers might be either. |
| What does a provider hand back? | Credentials and an address for a store; a command for a terminal. Same relation, different shape crossing it. |
| Is provisioning one mechanism or two? | The mesh's own registry is provisioned **before there is a mesh**, so provisioning is part of the bootstrap and part of what the carried bundle expresses. At bootstrap the store is local; afterwards it is on another node. Same operation, both sides of a tier boundary. |
| Is a tool surface one relation with two audiences, or two? | 56 of 126 modules carry tools — more than carry a service — and what consumes them is an **agent**, not a module. |
| ~~Do the remaining cross-context reads want an interface or events?~~ | **Derived, not chosen.** Neither is SQL — that only ever runs against your own store. ADR 0004 makes disconnection ordinary, so anything that must work while disconnected cannot use a request and needs a local copy: a subscription. Anything where a stale answer is worse than none cannot use a subscription. |
| What does a consumer do about events it missed while disconnected? | Replay from a point, ask once for a full picture and resume, or rebuild. The question every projection has, and unanswered here. |
| What happens to a grant when its consumer is removed? | Dropping is data loss; keeping is a leak. ADR 0005's removal rule does not obviously carry, because the thing lives inside another module's state. |
| Is a declaration composed per node, from what that node reported? | Some configuration follows the hardware. Either the host fills a blank — deciding, against ADR 0005 — or the control plane composes from the node's inventory first. |
| Would `excludes` and capability requirements actually be used? | Zero manifests declare either, which is equally consistent with *nobody needs them* and *nobody can express them*. |
| What does an exclusion mean for something already installed? | Refuse the install, or surface the conflict and let it be decided. |
| Are tiers a view of the graph, or a constraint on it? | If a tier is a computed level the word is a convenience. If *a tier may depend only on tiers below it* is to be enforced, it is a constraint and must be stated as one. |
| Where does resolution happen — the mesh, or the platform's package manager? | The mesh must model mesh-level edges. Whether it also resolves operating-system packages decides whether a solver gets written. |
| How far may the control plane be split? | A single board over several contexts works because they are contexts *inside* one control plane with one interface. If a context becomes its own deployable with its own interface, the board is coupled to N of them and the composition has nowhere to live that tier 3 permits. A constraint on splitting, worth knowing before splitting. |
| What does the pipeline schedule, once features are gone? | It schedules features today. The four categories they split into have different lifecycles, so the unit of work differs for each and needs naming. |
| Should placement leave the catalogue? | A provision pins itself to a named node in the manifest. Placement is an inventory decision, and having it in the catalogue means a second node cannot provide the mesh's store without editing its consumer. |
One module, worked out completely, because it is the case that breaks the tidy version. It
looks like *a container that runs a database* and it is at least nine things.
Postgres is the example. **The shape is not specific to it** — see the end.
## What it carries
**1 — A supervised container.** An image, a version, a data volume, and configuration. Easy,
and the only part the phrase "a docker service" describes.
**2 — Persistent state, and where it lives matters.** The volume is the database. Moving this
module between nodes is not rescheduling; it is a migration. Almost nothing else in the
catalogue has this property, and nothing in `provides` / `requires` expresses it.
**3 — Configuration that is partly the machine's.** Tuning follows the hardware — memory,
storage. A declaration generated centrally cannot know those, and
[research 012](../012-the-minimum-viable-node/00-overview.md) says the machine's own values win
on conflict. So some of this module's configuration is *derived from the node it lands on*.
**4 — A tool surface.** It exposes capabilities agents can call — query, list, provision. That
is not a resource on a machine and not an artifact; it is a contract the mesh publishes on the
module's behalf.
**5 — A provisioner.** The part that matters, and the one below.
**6 — Its own bookkeeping.** The provisioner must remember what it granted to whom, or it
cannot revoke, rotate or clean up. So a module that provides state to others *also* holds state
about its providing — and that state is not the database's data.
**7 — An exposure decision, per node it runs on.** Reachable from the machine only, from the
local network, or publicly. That is a property of *this assignment*, not of the module: the
same module on two nodes may answer differently.
**8 — Credentials it generates.** Per consumer, and they have to reach the consumer. Which
means a provisioning edge carries a payload, and the payload is a secret.
**9 — Health that is not "the container is up".** A container running and a database accepting
connections are different facts, and the second is the one anything cares about. This is the
host's read-back rule, at a distance.
## The provisioner is a second kind of edge
The tidy version of this effort said: *a module provides names, a module requires names, that is
the only edge.* A small game wanting to store data shows it is not.
```
my-cool-game requires postgres # I speak its protocol, it must exist
my-cool-game requires a database FROM postgres, called my-cool-game
```
The first is **presence**: the thing exists and is reachable. Nothing is created; nothing flows
back. `vscode requires terminal` is this, and so is `requires container-runtime`.
The second is **instantiation**: the provider is asked to make something *for this consumer*,
and hands back what the consumer needs to use it. A database, a user, a password, an address.
They differ in every way that matters:
| | presence | instantiation |
|---|---|---|
| creates something | no | yes, one per consumer |
| carries a payload back | no | credentials, an address |
| can be revoked | — | yes, and must be when the consumer goes |
| provider holds state about it | no | yes — who was granted what |
| satisfied by | anything providing the name | that provider, specifically |
**This is the mesh's actual power**, in the operator's words: a small game declares it wants a
database and the mesh makes one. Nobody creates a user by hand, nobody pastes a connection
string. That is worth being precise about rather than folding into a single relation because
one relation is prettier.
## What that costs the design
**The proposal's "one kind of edge" is wrong**, and the current system already knew: it has
`dependencies` for presence and `requires: provision:` for instantiation, with the resolver
deriving a presence edge from every instantiation edge. [`analysis.md`](analysis.md) recorded
that derivation as a convenience. It is not — it is the correct relationship between two
genuinely different relations.
So: **two kinds of edge, one graph.** Instantiation implies presence. Presence does not imply
instantiation.
## Which provider, when two nodes run one
Asked directly, because two nodes can each run a relational store and a consumer has to be
served by one of them. Neither obvious answer is right.
**Not "the consumer names the node."** That is placement in the consumer's manifest — a small
game edited because a database moved, which is the fault
[`proposal.md`](proposal.md) separates constraints from placement to avoid.
**Not "the consumer does not care" either.** For presence it genuinely does not: a terminal is a
terminal. For instantiation it cares permanently, because the data lands in exactly one store
and the wrong choice is discovered long afterwards.
**What the consumer does know is the scope of its own need.** Not which node — how many of the
thing it wants, relative to itself:
| Scope | Means | Example |
|---|---|---|
| **shared** | one instance serves every instance of this consumer | the mesh's own registry: every node reads the same rows |
| **per instance** | each instance of this consumer gets its own | a local cache, a per-node queue |
That is a property of the consumer, expressible without naming anything. And it is the thing
that actually decides: a shared need cannot be satisfied by a provider each node runs
separately, and a per-instance need should not be satisfied by a shared one.
**Then the mesh binds, and the binding is written down.** Not recomputed: a resolver that
re-derives which store serves a consumer will one day derive a different answer and relocate a
database, so the binding is made once and changed only deliberately.
**Where it is written down follows the scope.** A shared grant belongs to the *module* and every
assignment of it references the same one — which is the answer for one module installed on two
nodes wanting one database between them. A per-instance grant belongs to the *assignment*. Same
relation, two homes, and which home is not a detail: it is what makes two instances share
something or not.
The pieces that follow, none of them settled here:
- **When several providers satisfy the scope**, something chooses — most plausibly locality,
preferring a provider on the same node. That is a default, and it must be overridable, because
the reason to override it is exactly the reason nobody anticipated it.
- **A binding is a thing that can be wrong.** Once recorded it can be inspected, and a consumer
bound to a store on a node that no longer exists is a question somebody can be asked rather
than a failure at connect time.
- **Moving a binding moves data.** Whatever the mechanism, changing it is a migration and not a
configuration change, and a design that lets it look like the latter will lose something.
## Provisioning is early, not late
An assumption worth killing: that provisioning is something the control plane does for
consumers once a mesh is running.
**The mesh's own registry is a provisioned database.** So is its virtual host on the broker.
Neither exists until something creates them, and nothing in the mesh works until they do. The
order is:
```
1 the store runs from the bundle the host carries
2 a database is created in it a provisioning step
3 the mesh's own schema is applied a migration, against that database
4 the control plane starts and only now is there a mesh
5 everything else is provisioned the ordinary path
```
Steps 2 and 3 happen **before there is a mesh to do them**. So provisioning is not a
control-plane service that consumers use; it is part of the bootstrap, and part of what the
carried bundle has to be able to express.
**Which strains what a declaration is.** [ADR 0005](../../02-DECISIONS/0005-the-node-host.md)
has the host applying *declared state on this machine*. A database inside a running store is not
a file or a unit — and at bootstrap it is, at least, local: the store is on the same machine as
the host applying the bundle.
Later it is not. A consumer on one node provisioned from a store on another is the ordinary
case, and reaching it is not the host's job.
**Resolved as two mechanisms, which is the answer rather than a compromise**
([ADR 0005](../../02-DECISIONS/0005-the-node-host.md)). The host
runs bootstrap actions locally from the bundle; the control plane provisions across the mesh
afterwards. Different actors, different scopes, different trust paths — so there is no single
operation with a tier boundary running through it.
## Several modules, one database — and the case for refusing
Asked, then reconsidered by the operator: *maybe we should not allow it.*
The permissive version was a per-consumer **schema** inside a shared database — its own
namespace, its own migrations, revocable by dropping the schema, with a cross-context join
possible but deliberate.
**The stricter version is better, and it goes further than schemas.**
> **A module is only ever granted a resource it exclusively owns.**
No shared writes. And **no read-only role on another module's database either** — reading
another context's tables couples you to its layout exactly as firmly as writing them does, and
the coupling is harder to see because nothing breaks until the owner changes a column.
That is what [`how-we-build`](../../00-META/how-we-build.md) §4 already says: *contexts integrate
through the record, never through a shared schema.* The permissive version kept the letter of it
and left the temptation in place, and the path of least resistance wins eventually. A boundary
that is merely inconvenient to cross is a boundary that gets crossed.
### What it costs
**Cross-module reporting.** Anything wanting to know what several modules hold can no longer
join across them. It consumes their events, or calls their interface, and neither is as
immediate as a query.
That cost is the point rather than a regrettable side effect — it is §4's whole argument, and
the mesh already has both mechanisms: an event stream every context publishes to, and a tool
surface every module exposes. What gets harder is the thing that was making work belonging to
one context keep having to be implemented in another.
**One more connection per consumer.** A dozen modules means a dozen databases rather than a
dozen schemas in one. For a relational store this is unremarkable; it is worth stating only so
nobody discovers it as a surprise.
### What it deletes
The effort has been looking for what the design *removes* rather than adds, and this is the
first clear instance:
- **Grant kinds.** There is one — an exclusive resource. No schema grants, no read roles, no
scoping rules for who may see what inside a shared thing.
- **The question of who owns which table**, and with it the guessing at revocation time.
- **Cross-module migration ordering.** Two modules migrating one database need their migrations
ordered against each other. Exclusive ownership means a module's migrations are ordered only
against itself.
- **A whole class of permission modelling** that a shared store would otherwise need.
### The rule is about contexts, not processes
An earlier version of this file argued that a dashboard reading a dozen stores was caught by the
rule, because a dashboard is a surface and surfaces speak to an interface. **That was wrong, and
it drew the line in the wrong place.**
The mesh's own board showing nodes, modules and deployments is not a separate context reaching
across a boundary. It is the mesh showing its own data. Reading that store is not a violation
however it is done, and requiring it to go through an interface to reach facts its own context
owns would be ceremony.
**The rule is: a context is granted what it exclusively owns.** Everything inside that context —
its service, its surface, its tools — reads it freely. What is forbidden is a *different*
context reading it.
Which is exactly what the count below shows: the problem was never surfaces. It was three other
contexts keeping their tables in the mesh's database.
**And one surface over several contexts is normal.** The board visualises the mesh, the work
engine, the knowledge base and more, and the alternative — a separate web application per
context — is worse for everyone who uses it. That is not a compromise with the rule; composing
several sources into one view is what a surface *is*.
What it changes is only where it reads from: each context's **interface**, not each context's
**store**. Most of that already exists — 56 of 126 modules carry a tool surface, more than carry
a service.
**And the unified board is what keeps those interfaces honest.** If a view cannot be built from
a context's interface, that interface is inadequate — discovered in the one place where it is
cheap to notice, rather than the first time something else needs the same data and quietly
reaches for the store instead.
**But "the board calls each context's interface" is not quite it either**, and the objection is
right: if every context runs its own service with its own interface, the board is coupled to N
of them instead of N schemas, something has to compose them, and composition is logic — which
tier 3 says a surface does not hold. That moves the problem up a layer rather than solving it.
**The skeleton already answers this and the argument above talked past it.** `work` and
`knowledge` are not separate services; they are **contexts inside the control plane**, alongside
the record, inventory, delivery and the rest — and `api` is listed there as *the one interface
every surface speaks to*.
So the board speaks to **one** interface. Behind it the contexts stay separate: separate stores,
integrating through the record. But they are one tier, one repository, one deployable, and
coupling *within* a tier is not what the tier rule forbids.
Which resolves the objection rather than deflecting it: the problem does move up a layer, and
the layer it moves to already exists and has this as its job.
**The caveat is load-bearing.** This holds only while the contexts are not separate deployables.
The moment one becomes its own service with its own interface, the board is back to N clients,
something must compose them, and the composition has nowhere to live that tier 3 permits. That
is a real constraint on how far the control plane may be split, and it is worth knowing now
rather than discovering it by splitting.
If composing even one interface turns out too slow, the answer is a projection the board owns
and keeps current from events — not access to somebody else's tables.
### Checked against the real consumers
The rule's survival turned on whether every reader of the mesh's own registry could be served
some other way. **Eighteen consumers open a direct connection to it.** Four groups, and only one
of them is work.
| Group | What happens under the rule |
|---|---|
| **The owner and its machinery** — the mesh module, the SDK, the environment and configuration synchronisers, secrets | Nothing. It owns the database. |
| **Node appliers** — the overlay, the shell daemon, the resolver | **Already resolved.** [ADR 0005](../../02-DECISIONS/0005-the-node-host.md) stops the host querying the mesh database, decided for tier reasons with nothing to do with this. |
| **Foreign tenants** — the work engine (10 tables), the knowledge base (2), pipeline logs (1) | They need **their own database**. They are not reading the registry; they are storing their own data in it. |
| **Genuine cross-context reads** — the work engine reads `nodes`; two others read a handful | The only ones needing an interface or events. |
**Thirteen foreign tables live in the mesh's registry database**, belonging to three separate
contexts. That is [`how-we-build`](../../00-META/how-we-build.md) §4's shared schema, counted.
**So the question was the wrong shape.** The bulk of the problem is not readers needing a new
route to data — it is **tenants needing to move out**. Tasks, agents and teams have nothing to do
with nodes and modules; they are co-located by history. Give that context its own database and
its dependency on the registry shrinks to a single table.
What remains is a handful of genuine cross-context reads, small enough to enumerate rather than
estimate. **The rule holds.**
### Request or subscription, and what decides
The remaining cross-context reads need one or the other. **Neither is SQL** — under exclusive
ownership a module runs SQL against its own database and nothing else, whatever transport a
query might travel over. Both options are the mesh's own channel, and both ride the broker
([ADR 0002](../../02-DECISIONS/0002-nodes-communicate-over-a-broker.md)), so the transport is
not the distinction.
**The distinction is where the answer lives when you need it.**
| | request | subscription |
|---|---|---|
| you ask | at the moment you need to know | never — you are told |
| the answer lives | on the other side | in your own store |
| freshness | always current | as current as the last event you received |
| when the other side is down | you cannot answer | you answer from your copy |
| what you must handle | a round trip that can fail | events you missed while you were down |
**What decides is not taste.** [ADR 0004](../../02-DECISIONS/0004-a-node-and-how-it-joins.md)
makes disconnection an ordinary situation rather than an exception. So:
> **Anything that must keep working while disconnected cannot use a request** — there is nobody
> to ask. It needs a local copy, which means a subscription.
And the converse: anything that must be *correct at the instant of asking*, where a stale answer
is worse than no answer, cannot use a subscription. A display can lag. A decision about whether
a grant is still valid cannot.
That turns an apparently open question into a derived one. What remains genuinely open is
narrower: **what a consumer does about the events it missed** while it was disconnected — replay
from a point, ask once for a full picture and resume, or rebuild from scratch. That is the same
question every projection has, and nothing in the record answers it yet.
## A migration belongs to the consumer and runs on the provider
A game defines migrations. They run against the database the store granted **it**. So the
migration is:
- **owned** by the consumer — it is that module's schema, versioned with that module;
- **hosted** by the provider — it runs inside something the consumer does not control;
- **ordered** after the provisioning edge — there is nothing to migrate until the grant exists;
- **scoped** to the grant — the consumer's migrations touch its database and no other.
Ownership crosses the edge, which nothing in *provides* and *requires* expresses. And it is the
*action* category from [`features.md`](features.md) made concrete: not an artifact, not node
state, and not something the host can apply, because the thing it changes is not the machine.
It also gives a consumer's own install an internal order — **provisioned, then migrated, then
started** — that depends on an edge rather than on the module's contents.
## What still has no answer
**How many instances of postgres should exist?** One per mesh is wrong — a node that must work
while disconnected cannot depend on a database elsewhere. One per node is wrong — the mesh's own
registry is one thing, not one per node. So the answer is per-module, and nothing in the schema
says it. This is [`cases.md`](cases.md) axis *how many instances*, and postgres is the case that
proves it cannot be a global rule.
**What happens to a grant when the consumer is removed?** The game is uninstalled. Its database
still exists, holding its data. Dropping it silently is data loss; keeping it forever is a leak.
arriving from a direction that issue did not anticipate. And a fleet where every node kept its
own settings is a fleet where a module works on one node and fails on another with nothing in
the mesh able to say why.
**The distinction that dissolves both rules.** Neither direction is right as a blanket, because
the question is not *whose configuration wins*. It is whether the module **requires** the
setting or merely **prefers** it — required contradictions cannot be kept without breaking the
module, and preferences should always yield to what is already there.
That is a property of the module's own declaration rather than of the adoption algorithm, which
makes it one more thing the graph would carry
([research 011](../011-the-module-graph/00-overview.md)). Until modules can say which of their
settings are load-bearing, adoption is choosing a default in the dark, and the default chosen
here is the one that does not break the machine it is adopting.
### The briefing carries an outcome, and the outcome is derived
Proposed by the operator: the report states plainly whether adoption succeeded, partly
succeeded or failed, and each line carries its own severity.
**Each line is marked, and the overall is the worst mark present.** Derived rather than stated
alongside, because two fields written independently drift — and a briefing reading *full
success* while carrying a failed line is exactly the fault this record keeps cataloguing. An
outcome computed from its lines cannot disagree with them.
| Mark | Means |
|---|---|
| `ok` | done, and verified |
| `kept` | a disagreement; the machine's value was kept and somebody should look |
| `unknown` | could not be determined |
| `failed` | could not be done — the node is not what was asked for |
**`unknown` is not a shade of success.** Adoption will meet configuration it cannot parse and
state it cannot read, and folding those into *fine* is the same move as reporting an installed
package as a capability. A thing nobody could determine is a thing nobody can rely on, and it
gets its own mark for the same reason a capability detector reports *why*.
**And this opens something the earlier rule did not cover.** *Flags inform, they do not block*
was decided about **conflicts** — where the mesh chose, deliberately, and the machine still
works. A **failure** is different in kind: not *we chose* but *we could not*. Treating both the
same makes a node where something the mesh needed never happened indistinguishable from one
where a log level differed. Whether a failed line still lets adoption complete is therefore
reopened by adding severity, and is not decided here.
## Open questions
| Question | Why it is open |
|---|---|
| What is the closure for a one-node mesh? | The skeleton asserts four pinned services. [Research 006](../006-mesh-from-scratch/00-overview.md) already asks whether it is four or five and does not answer. A graph gives a computed answer instead of an asserted one, which is [research 011](../011-the-module-graph/00-overview.md). |
| Is "tier" the same thing as a graph level? | Tiers were named as a bootstrap order. If the closure is computed, tiers may be a derived view of the graph rather than a separate concept — or they may be a coarser boundary that survives for a different reason. |
| ~~What happens when existing configuration contradicts what the mesh needs?~~ | **Answered** — the machine's configuration is kept, the conflict is flagged, and it is reconciled afterwards. |
| ~~Do flags block, or only inform?~~ | **Answered** — they inform. Adoption always completes, and the node is a node. |
| Can a module say which of its settings are load-bearing? | The question that dissolves the conflict rule rather than choosing a side. A setting the module *requires* cannot be kept from the machine without producing something installed and broken; a setting it merely *prefers* should always yield. Until a module can say which is which, adoption is defaulting in the dark. Belongs with the graph. |
| Does a `failed` line still let adoption complete? | *Flags inform, they do not block* was decided about conflicts, where the mesh chose and the machine works. A failure is *we could not*, which is different in kind — and treating them alike hides the worse one behind the commoner one. |
| How is a flagged conflict reconciled, and by whom? | The briefing hands it to a session. What that session is empowered to change, and whether the resolution is recorded so the next adoption does not re-raise it, is undecided. |
| Where does the kept original live, and for how long? | Whether it is recorded in the node's state so adoption is visibly reversible, and whether it is returned when the mesh stops managing the thing. |
| What shape is a briefing? | Structured enough to be acted on, prose enough to be read. It is the first thing a session on a new node sees, which makes it an interface rather than a log. |
| Does owning a package mean owning its version? | Owning configuration and owning the package are different scopes. The second means the mesh decides which version is installed, and that decision then has to survive the machine's own package manager updating it. |
| How does a bundle stay true between building and applying? | It is built against a scan of the target. The machine can move between the scan and the apply, so the host has to verify rather than assume — and fail plainly when the bundle no longer fits. |
| What cannot be precomputed at all? | Anything built from source on the target still needs a toolchain and a network at that moment. Tailoring moves that cost rather than removing it, and *minimal viable* has to be honest about what it cannot ship ahead. |
| Does presence differencing understate the gap? | Knowing a package manager is installed does not say it is the version the mesh needs. A difference computed on presence alone is optimistic. |
# 1. The mesh brokers capabilities; nodes host; agents think
## Context
HAL has 124 modules. The count is not the problem — it is the symptom. Modules are split
because splitting is the only granularity the platform offers, and domains are merged
because a shared database is the only integration it offers. Both pressures push in the
same direction: boundaries end up drawn by deployment accident rather than by domain.
Three observations establish the state.
**The word "agent" means two different things.** The original design treated a node as an
agent with thinking abilities. Later, noxflow implemented agents as employees with
skills, workspaces and tasks. Both survive. The collision is visible in the data — there
are **two agent rows per node**:
| agent | skills |
|---|---|
| one named after the node | `{deploy,verify,operate}` |
| one named `hal-<node>` | `{}` |
One carries the work; the other carries only identity, existing to hold a licence for
HAL's own sessions. The same split appears in the schema: `nodes.hal_claude_account` and
`agents.claude_account` are one fact in two tables in two databases, and what was
documented as "three licence touchpoints" is one concept modelled three times.
**Domains integrate by sharing a schema.** `noxflow` is 45 tables spanning five domains —
tasks (7), agents (13), knowledge (8), meetings (7), scheduling (2). Its original purpose
is 7 of 45. Because agent identity lives in noxflow's database, work that belongs
elsewhere must be implemented there: per-agent Claude credentials had to be written by
the noxflow runtime, even though node identity — the same kind of fact — lives in the
mesh registry.
**The core domain has no context to live in.** The mesh's distinguishing feature is that
a module declares `requires: postgres/database` and never learns where the database
lives, who owns the credential, or how it rotates. That is capability brokering, and it
is what makes this a mesh rather than four machines with a configuration manager. Yet the
logic implementing it sits in `modules/postgres/tools/index.ts` — a provider module,
where "the mesh brokers credentials" cannot be expressed. On 2026-08-22 three of its
invariants were found violated simultaneously (see Consequences).
## Considered Options
1. **Keep the current layout; fix bugs as they surface.** Rejected. The faults are not
independent. Every incident on 2026-08-22 — credentials written by the wrong module, a
dependency question answered wrongly twice, an invariant enforced nowhere — traced to
a boundary that was never stated. Fixing them individually leaves the generator intact.
2. **Merge aggressively into few large modules.** Fewer names, same problem: a shared
schema across domains is what produced noxflow, and doing it deliberately would
produce it again at larger scale.
3. **Decompose by bounded context, with the mesh as a broker.** Name contexts after their
aggregates, integrate through a published record rather than a shared schema, and let
deployment granularity be a feature-level concern rather than a reason to create a
module. **Adopted.**
## Decision
### The domain, in one sentence
**The mesh brokers capabilities. Nodes are places where work runs. Agents are personas
that think and act.** Everything else supports one of those three.
### What this is, plainly — and what "mesh" does not mean
*Written 2026-08-29, from working through connectivity and asking whether the word still fits.*
Four layers. Naming them honestly is worth more than the word on the tin:
| | |
|---|---|
| **machines are linked by a private network** | and every machine reaches every other over it |
| **one node holds knowledge of all of them** | the control plane, and only it |
| **modules are how anything is built and delivered** | this *is* the CI/CD, not something beside it ([ADR 0010](0010-delivery.md)) |
| **every node runs a session you can message** | a feature of the node, remembering across callers; any node can message any other ([ADR 0004](0004-a-node-and-how-it-joins.md)) |
| **workers are hired onto nodes to do tasks** | employees, with a lifecycle — a different thing from the row above ([ADR 0003](0003-agents-are-persistent-employees.md)) |
**The last two rows are not the same thing and the vocabulary of one does not describe the other.**
A node's session comes with the machine: nobody hires it, it holds no tasks, it is never
reassigned, and it goes when the node leaves. A worker is an employee — named, hired, drained,
retired, movable. They are built from the same parts and run on entirely different terms, and
collapsing them is how the employee vocabulary ends up stretched over something it does not fit.
**Neither makes the node itself a thinking thing.** A node is a machine; both of these run *on*
one, which is why *a node does not authenticate to a model provider, agents do* is unaffected by
either.
**Where the value is** is that both can reach across the whole set: a shell, a service, a file, or
simply a question to another node. Not machines that can be configured centrally, which is
ordinary, but a set of machines that can be worked across as though they were one.
**This is not a mesh in the peer-to-peer sense and will not become one.** The word describes what
machines can reach, not how they are governed:
| | a mesh? |
|---|---|
| what a machine can reach | **yes** — genuinely any to any |
| how the traffic travels | no — anything crossing sites transits the hub |
| who decides | no. One node, declared |
**And *master* overstates it in the other direction.** A master implies the others need it in
order to function. They do not: every node holds what it was last told and runs from that copy
*always* — not as a fallback, as the only mode it has. So the control plane being gone is every
node in the ordinary disconnected situation at once, and **what is lost is change, not
operation.**
The accurate phrase is **one authority, no failover**, and both halves are deliberate
*Consolidated 2026-08-28 from three records. Overlay, resolution, exposure, filtering and
certificates are one design.*
## Why it is control-plane work
Apply the test — *everything that needs to know about more than one node* — and not one of the
five can be answered by a machine on its own:
| | needs to know |
|---|---|
| **overlay** — who peers with whom | every node, and which can be dialled |
| **resolution** — which name is which node | every node |
| **exposure** — which public name reaches which container | which node is publicly reachable |
| **filtering** — which port is open, to whom | what is assigned here, and the overlay's shape |
| **certificates** — who may present which name | which name belongs to which node |
That is exactly what the current arrangement gets wrong, by computing all five on the node from a
direct database connection. Two modules do this, and they are the only two left holding a
credential to the control plane's database.
**The shape of the fix, once for all five:** the connectivity context computes the configuration;
it arrives over the link as `file` resources; the service reads files and knows nothing about the
mesh. **This costs no new host vocabulary.**
## A route is a grant
**Ingress is not substrate.** The control plane does not need a route to start — it listens
locally — and no node needs one to reach it, because the node dials out and has no listening
control surface. It grants itself a route afterwards, the way it grants itself a bucket.
The strongest objection deserves stating: the `api` is the one interface every surface speaks to,
so eventually it *does* want a public name. But **wanting one later is not needing one to
start**, and that distinction is the entire substrate test.
**A module that must be reachable declares it needs a route; the proxy provides one.** Ordinary
instantiation, with the direction mirrored — the consumer supplies a target and receives a name.
**Exposure is three facts at two scopes**, which is why it cannot live on the node:
| the fact | scope |
|---|---|
| the public name resolves to an address | **mesh** — which node is publicly reachable |
| a certificate valid for that name exists | **mesh** — issued once, used on one node |
| the proxy maps that name to that container | **node** |
**A node without a public address is proxied by one that has**, across the overlay. Most nodes sit
behind a connection with no forwarded port, so exposure cannot assume the workload's node is
reachable.
## Reachability is declared, not inferred
The overlay's peer graph is computed from whether a node can be dialled, and that was inferred
from a regular expression over the address. **The address is evidence of reachability; it is not
the fact**, and the gap has already cost:
| address | the regex says | actually |
|---|---|---|
| `100.64.0.0/10` — carrier-grade NAT | **public** | **not reachable.** An endpoint is written to an address nothing can reach |
| any IPv6 address | public | the test is v4 shapes only |
| a routable address behind a closed firewall | public | not reachable |
| a documentation range standing in for a public segment | private | reachable — this is the lab bug |
**A test environment having to choose its addresses to satisfy a regex is the regex telling us it
is not a fact.**
So: **an endpoint, or none** — declared. And **the hub is declared, never derived from an address
prefix**, because an election decided by the first four characters of an address fails silently,
cannot be queried, and makes a renumbering an outage.
**The address remains evidence and stops being the fact.** Where an observed endpoint disagrees
with a declared one, the disagreement is a **reportable condition**, not a silent correction.
**What does not change** is the lesson underneath: role does not imply reachability — a
home-hosted node is a server that cannot be dialled. This keeps that and stops encoding it as a
pattern match.
### Some nodes must be reachable, and this had not been said
*Written 2026-08-29, on being asked and finding no answer.*
Everything above treats reachability as a **fact to record** — which node can be dialled, so that
exposure and certificate issuance can be placed. It never said the converse, and the converse is a
hard requirement:
| role | dialled by | so it needs |
|---|---|---|
| **the node running the broker** | every node, outbound ([ADR 0002](0002-nodes-communicate-over-a-broker.md)) | to be reachable from wherever nodes are, at a **stable address** |
| **the hub** | every node not co-located with its peer | the same |
**Across the internet, "reachable from wherever nodes are" means publicly reachable.** For a mesh
confined to one network it does not — the requirement is about the nodes that exist, not about the
public internet.
**Stable is the sharper half.** A token carries the broker's *address, not a name*, because there
is no resolution before joining ([ADR 0004](0004-a-node-and-how-it-joins.md)). A broker node whose
address moves invalidates every token issued for it, and a node that was disconnected across the
change cannot get back.
**A mesh whose nodes are all behind NAT cannot be raised.** That is a real precondition and it
belongs with the others rather than being discovered.
### The link stays on the underlay, and that is a repair channel
The obvious objection is that all traffic should run over the overlay. Nearly all of it does — SSH,
services, node to node — and the exception is each node's own outbound link to the broker.
**At join time it is forced**: a node has no overlay yet, so it cannot use one to ask for one.
**Afterwards it is a choice**, and the reason is that the link is how a broken node is fixed. A
repair channel carried over the thing being repaired is not a repair channel: a node whose only
path home was the overlay is gone the moment an overlay declaration is wrong.
**What it does not cost is confidentiality.** The link is already authenticated and encrypted
against a pinned fingerprint ([ADR 0004](0004-a-node-and-how-it-joins.md)), so moving it onto the
overlay would not protect traffic that is unprotected today.
## A filter rule names its source
`scope: public` is declared in five manifests, is part of no rule type, and is **referenced by no
code**. So five manifests appear to restrict a port and restrict nothing — on the modules most
worth restricting.
**A rule names its source. `from:` is the only way to scope one, and a rule without one is open**
— which it must say plainly rather than appear to deny.
**`scope:` is removed rather than implemented**, because giving it meaning would leave two ways to
express one thing. And the general fix is that **an unknown key is refused**: the host's
declaration parser already works this way, and manifests are the layer where that discipline is
missing. `scope:` survived because nothing rejected it, and it spread by copying to five
manifests.
## Order, and what it costs
**The link runs on the underlay and never on the overlay.** The overlay is configured by the mesh,
so a link requiring it could never be established on a new node.
**The first declaration is the overlay and nothing else** — because a node's address and peers are
*assigned* so it cannot come earlier, and because it is the way back in. A node reachable over the
overlay can be fixed by hand if a later declaration breaks the machine; **a large first
declaration risks a node that is broken and unreachable at once.**
**Reachable is not the same as having a control surface.** Every node reaches every other over the
overlay — SSH, services, ordinary traffic — and every node consumes from the broker. What is
forbidden is a listening thing that accepts instructions and changes the machine.
## Consequences
- **The last two direct database connections leave the nodes**, and with them the database
credential every node carries.
- **The `/etc/hosts` floor goes**, along with the bootstrap circularity it patched.
- **Two certificate authorities stay separate on purpose**: a public one for public names, the
mesh's own for internal ones. A single-CA lab would hide any bug living in the split.
- **What happens when the hub is down**: nothing takes over. Non-co-located paths stop; co-located
peers and every assigned workload keep running.
## Open — the link over the overlay, with a fallback
*Raised 2026-08-29 and deliberately left open, because the honest gain is smaller than it looks
and it is a decision rather than a derivation.*
The proposal: a node prefers the overlay for its link and drops to the underlay when the overlay
is not working — so ordinary operation is private and the underlay stays as the way back.
**Two things it would have to get right:**
- **The trigger cannot be "is the overlay up".** A WireGuard interface has no link state; once
configured it is up whether or not the far end exists. So there is no flag to read, and failing
over means *try, fail, time out, retry elsewhere*.
- **Running on the fallback has to be visible.** A node that quietly drops to the underlay is a
node whose overlay is broken with nothing to say so, and it will stay broken because everything
still works. That is this repository's recurring fault — a failure that reads as success — and a
fallback is the easiest place in the design to reintroduce it.
**What stops it being an obvious win:** if the underlay path must stay available for the fallback,
the broker stays exposed on it. So the exposure is unchanged and the traffic was already encrypted
— the gain is which network carries bytes, not what an attacker can reach.
**The version that would buy something is overlay-only**, with the broker firewalled to the overlay
in steady state, accepting that a node whose overlay breaks needs hands-on recovery. That is a real
trade: it exchanges the automatic way back in for a closed port.
One kind of thing, one manifest describing all of them. A database, a web application, a window
manager and a firewall rule set are all modules — not because they are alike, but because
**anything else means a second kind of thing with its own rules, and then a third.**
**A module is the unit of delivery**: assignable to a node, versionable, replaceable on its own.
## There are no domain modules
An earlier decision grouped modules by domain — four things constituting *how a node is
reachable* becoming one `networking` module. **That was wrong, and the correction is worth
keeping** because the observation behind it was right.
The measurement holds: reachability is the **only** place in the catalogue where modules
genuinely change together under one intent. What did not hold is the conclusion. Tight coupling
means they share an **authority** — one place that decides for all of them — and not that they
should be one artifact. `wireguard` and the proxy are deployed to different sets of nodes, so a
module containing both would be assigned where half of it is unwanted.
> **Coherence is a context. Delivery is a module.**
**Folders assert relationships; edges record them.** What grouping was for — finding things,
seeing what belongs together — is a tag and a query, neither of which anybody has to keep true by
hand.
### What a domain module turns out to be, and why it is not the one refused above
*Written 2026-08-29, from building it. The heading above reads as a contradiction of what now
exists and is not one — but only if the difference is stated, so it is stated here.*
**What was refused contains things. What exists contains nothing.**
| | `networking` as refused | `networking` as built |
|---|---|---|
| what is in it | WireGuard, a proxy, a firewall — artifacts | nothing at all |
| what it says | *these ship together* | *I want a private network and names* |
| what is assigned | one module, half of it unwanted | whatever answers each requirement, each on its own |
The objection above is untouched by this and still correct: a module holding WireGuard and a
proxy is assigned where half of it is unwanted. **A module holding nothing cannot be, because
there is no half.** It is requirements and a name, and every artifact it leads to is still an
ordinary module assigned on its own terms.
**Why it is worth having.** Most people want the network working and do not want to choose a VPN.
`assign networking` finds one answer to each requirement and takes it without asking, because
with one candidate there was never a question — the rule below about refusing does the work.
Somebody who does care assigns the VPN they want, and *that is the whole of choosing*: there is no
flavor field, no variant syntax, and no second verb. **Picking an implementation is assigning a
module.**
**What it costs, stated because it is real.** Adding a second implementation to the catalogue
turns a settled question into an open one for **everyone using the bundle**, not only for whoever
wanted the alternative. Every node assigned `networking` refuses until somebody says which. That
is [the refusing rule](#a-requirement-with-several-answers-is-refused-never-guessed) applied
consistently, and the alternative is a default — which is the flavor field returning under a
better name. The cost is one assignment per node, and the message names the candidates.
**A consequence that had to be found by running it.** A bundle can drag an implementation in
through a requirement nobody looked at. Choosing a different VPN still installed WireGuard,
because the names module needed addresses only WireGuard hands out, and nobody was told. Two VPNs
on one machine is not always wrong — a machine may run one for another purpose — but being **the**
network the mesh runs over is singular, so that is a claim, and the collision is refused by name.
**The general rule: what a bundle pulls in is only as safe as the claims on what it pulls in
from.**
## Three edges
| edge | means | declared? | satisfied |
|---|---|---|---|
| **presence** | that thing must exist and be reachable here | yes | at provisioning |
| **instantiation** | that thing makes something for me and hands back credentials — a database, a bucket, a route | yes | at provisioning, and again whenever it must be |
| **build** | I was compiled against that artifact | **no — read from imports** | **at build, once** |
**Instantiation implies presence; presence does not imply instantiation.**
**A route is an instantiation edge**, and it is worth noticing because the direction is the mirror
of a database: the consumer supplies a target and receives a *name*, rather than supplying nothing
and receiving credentials. Same edge.
**Provider stops being a category.** Any hosted thing can be a factory — an identity provider
grants clients, a mail server grants mailboxes. It is a facet, not a kind.
**A module may also declare what it claims**, because some things cannot coexist and that is a
fact about the module rather than about a particular node. What that means precisely is below.
### Where the answer to a requirement is allowed to live
*Written 2026-08-29, from building it. The table above distinguishes **presence** from
**instantiation** and this is the half of that distinction nobody had noticed was missing: not
what the edge hands over, but **where the thing on the other end is.***
Two different things were both being written as a requirement:
| where the answer lives | **this machine** | **somewhere in the mesh** |
| how it is answered | install another module here | find the node already running it |
| what is missing if absent | a module to assign here | **a decision about where**, which is nobody's to make silently |
Answering the second like the first installs a database on every machine that uses one, which is
what it did.
**So a provided name carries a scope**, the same idea a claim already has, and written short in
the ordinary case so the few that are not node-scoped stand out rather than drowning. Scope is a
property of **the name, not of each provider**: two modules disagreeing about whether a database
is local would make one requirement mean two things depending on which happened to answer it, so
that is refused.
**A requirement answered from the mesh is never satisfied by installing it here.** Nothing, and
the mesh refuses and says which module to assign somewhere. Two, and it refuses and says how to
choose — the same rule as everywhere else, for the same reason: picking is guessing, and the wrong
guess puts somebody's data on a machine they did not choose.
**Choosing is recorded per node**, because that is the granularity the choice actually has — two
machines may reasonably use two different databases and a mesh-wide answer could not say so. A
choice pointing at a machine that does not provide the thing is refused rather than quietly
replaced by one that does, and a single available provider does not override a choice either.
**Both are the same rule: the mesh does not overrule a person, and it does not move data without
being told to.**
**What this is a prerequisite for.** Knowing *which node* answers is the first half of handing a
credential back — you cannot be given a database's password before it is settled whose database it
is. So a node's resolution now records what it takes from elsewhere, which is both the only part
of its set that stops working when a *different* machine goes away, and the place a credential
will hang.
### An edge has two directions, and only one of them is built
*Written 2026-08-29, from building it. The row above already says a consumer **supplies a target
and receives a name**; what it did not say is that those are two separate mechanisms, and that
having one without the other is what forced two modules outside the system entirely.*
| direction | the consumer says | who needs it |
|---|---|---|
| **contribution** | *publish me at this name, on this port* | the proxy, the DNS server, a firewall |
| **binding** | *and give me back a credential to it* | the database, the object store, the identity provider |
**Contribution is built.** A module declares what it contributes to a requirement; the control
plane collects every contribution on a node and writes them to a path the provider named, as a
file, in the mesh's own shape. **Contributing to something is requiring it** — asking to be
published means a publisher must exist, and a module that had to say both would eventually say
one, with the failure appearing as a machine where nothing serves the route.
**The control plane does not know what a reverse proxy is**, and does not write one's
configuration. It delivers the facts; the module turns them into whatever it runs. That boundary
is what makes swapping the proxy cost nothing in any module that publishes through it, and it is
[the same separation](0001-mesh-brokers-nodes-host-agents-think.md) that keeps third-party
software *on* the mesh rather than *of* it. It also costs the host nothing: a received file is a
file, which was checked by putting the control plane's output through the host's own parser rather
than by asserting it.
**Binding is built except for the secret**, and that turned out to be the useful way to cut it.
A provider says what a consumer needs in order to use it — a port, a driver, a realm — and a
consumer says where it wants to be told. The mesh adds the half only it has: **which machine, and
what that machine is called on the private network.** So an application on one node is handed the
address of its database on another, as a file, and reaches it by a name the mesh also created.
**The file states that it carries no credential, and why.** A missing field looks like a bug; a
stated absence looks like a boundary, and somebody wiring this up should not spend an afternoon
looking for a password that was never going to be there.
### And the secret, which is delivered without ever being held
*Written 2026-08-30, after looking at how the existing mesh does it. The design here is a reaction
to a measurement, not a preference.*
**The obvious arrangement is a credentials column, encrypted at rest.** It exists, and its own
tooling records what it bought:
| | |
|---|---|
| the tool for finding a secret matches **by value**, not by name | because one password is in the provisions table, the environment table, each node's environment file in plain text, and **inside every connection string composed from it** — copies its documentation calls *"often the only copies actually in use"* |
| a query against the encrypted column **returns zero rows and proves nothing** | so auditing moved to the decrypted copies on the machines |
**Two faults, and encryption at rest addresses neither.** The control plane can read what it
stores, so a copy of its database is a copy of every credential in the mesh. And one secret has
many homes with nothing tracking them — **composition is what mints the untracked ones**, because
building a connection string centrally creates a new secret-bearing value no rotation path knows
about.
**So the value is sealed to the node that will use it before it is stored.** With a key that node
generated and whose private half the mesh has never seen — a third key beside the identity and the
overlay, for the same reason those are two rather than one. What is stored is unusable by whoever
holds it, the mesh included, and the broker relays a blob it cannot read. This is what makes
[ADR 0004](0004-a-node-and-how-it-joins.md)'s *compromise of a node is compromise of that node*
true of secrets rather than true of identity and quietly false of everything that matters.
**And nothing is composed centrally.** A connection string is assembled on the machine that needs
one, if at all. The mesh delivers parts.
**What it costs, stated because it is real:** the mesh cannot audit by value. That is the right
trade rather than an oversight — a query over an encrypted column could not either, so the audit
was never real. What *is* answerable is which node holds what, which is the question rotation
actually asks.
**A consequence that shapes the mechanism.** The mesh discarded the plaintext, so it cannot
compose a file containing it. The credential is therefore **its own file**, holding the value and
nothing else, beside the readable one. That is better than the alternative it was forced into:
the readable half stays readable in the declaration, and the secret half changes only when the
secret does, so a service reloading on it reloads for a real reason.
**Rotation is generating a new one**, because reading the old one back is not possible. Both ends
are re-sealed and reach their machines in the same push — which removes the window where half the
mesh holds a dead credential, the failure
[recorded in ADR 0001](0001-mesh-brokers-nodes-host-agents-think.md) as consumers on three nodes
holding one for two days.
### The provisioner, which is where the mesh stops
*Written 2026-08-30, from building one and running it against a real database.*
**A password nothing was told to create authenticates nowhere.** The mesh generates one, seals it
to both ends and cannot read it — so it cannot tell the software to start accepting it either.
Something on the providing machine reads what arrived and makes it true. That is a provisioner.
**It belongs to the module, not to the mesh**, and the boundary is the same one that keeps
third-party software running *on* the mesh rather than being *of* it
([ADR 0001](0001-mesh-brokers-nodes-host-agents-think.md)). The control plane decides and never
touches a machine. **What the mesh owns is the contract**, which is two files the host writes from
an ordinary declaration:
| | |
|---|---|
| the manifest | every consumer, what it asked for, and **where** its credential is |
| one file per consumer | that credential, alone in it |
Two files because the mesh discarded the value and cannot compose a document containing it. As
before, the constraint produces the better shape: the readable half stays readable and auditable
in the declaration, and the secret half changes only when the secret does.
**It reconciles; it is never told what changed.** It runs after every declaration and must reach
the same state from wherever it starts. Three consequences, and each of them is a fault that has
been shipped somewhere:
- **the password is set every time, not only on creation** — otherwise the role already exists,
nothing happens, and a rotation reports success while changing nothing
- **what it made and nobody asks for any more is removed** — otherwise a consumer that left keeps
a working login for ever and nothing ever says so. This is the same rule the host follows about
[removing what it declared and no longer declares](../04-ISSUES/010-the-first-declaration-destroys-the-substrate/00-report.md)
- **what it did not make is left alone** — otherwise it cannot be run on a system that predates
it, which is every system anybody would want to adopt
**A missing credential is refused rather than worked around.** A role created without one is a
login nothing can use, and nothing would report it until something tried to connect.
**This is where the mesh stops**, and saying so is the point of the section. It decides, delivers
and can prove what it delivered; the last inch belongs to whoever knows what `create role` means.
**One check that only became possible now.** Two machines wired together across no private network
is a mesh that reports itself configured and does not work, and the failure surfaces as a
connection timing out — the slowest place to find anything. It is refused, and it is only
*checkable* because the network became [something a machine is
given](#what-a-domain-module-turns-out-to-be-and-why-it-is-not-the-one-refused-above) rather than
something it has by virtue of holding an address.
**What the absence cost, measured.** Exactly two modules opened a direct connection to the control
plane's database — the proxy and the VPN — and they are the reason every node permanently holds a
credential to it. Both were doing by hand what this edge is for. The VPN's half is closed by being
[a module whose files are computed](../03-DESIGN/01-to-be/08-connectivity.md); the proxy's is
closed by contribution. **Neither needed a new kind of thing, and both had been outside the model
for as long as there was one.**
### Why the build edge is a different kind
It is fixed inside an artifact rather than negotiated when something runs, and **its only remedy
is a rebuild** — nothing can re-provision it.
It is also **derived rather than declared**, and the asymmetry is deliberate: a runtime edge is an
*intention* somebody has about how the mesh should be wired, and only a person can state it. A
build edge is a *fact about code that already exists*, and a declared list of dependencies drifts
from the imports it describes.
**An artifact is out of date when its source moved, or when anything it was built against moved.**
So what is recorded is a commit *and the identity of every artifact it was built against*, which
is what makes the rebuild set computable and *is this current?* answerable without building.
**The graph measures design quality, not just build order.** A module with many inbound build
edges is one whose every change is expensive — and that is readable before anything is built. The
current shared library is exactly that, and nobody could see it because nothing drew the edges.
## Provisioning is declared, never configured by hand
A module declares what it **provides** and what it **requires**. The mesh satisfies it: a
provisioner belonging to the provider creates the resource and its credential, records the grant,
and the values are derived onto the consumer. **Neither the credential nor the topology is ever
written by hand.** A requirement may name a provider on another node, so cross-node wiring is the
same declaration.
## What a module claims, and why it is not a list of rivals
@@ -15,11 +15,15 @@ Where the two disagree, the implementation wins and the disagreement is stated.
| [`04-delivery.md`](04-delivery.md) | Push to running: the three silos, levels, and what a green pipeline proves |
| [`05-runtime-and-installation.md`](05-runtime-and-installation.md) | The node runtime, its modes, and how a node comes into being |
| [`06-configuration-and-secrets.md`](06-configuration-and-secrets.md) | Managed files, value resolution, and where secrets live |
| [`07-knowledge.md`](07-knowledge.md) | The two knowledge stores, and what each is for |
| [`07-knowledge.md`](07-knowledge.md) | The mesh keeps no store: what it knows is what modules answer, and the record is read by one |
| [`08-agents-and-work.md`](08-agents-and-work.md) | Agents as employees, tasks, workflows, and the meeting model |
| [`09-interfaces-and-observability.md`](09-interfaces-and-observability.md) | How the mesh is reached and watched — tools, board, proxy, health, thoughts |
| [`10-module-catalogue.md`](10-module-catalogue.md) | The catalogue's shape, and what its shape says |
| [`11-the-lab.md`](11-the-lab.md) | The lab — the first piece of the new shape that exists, and what it does not yet do |
| [`12-the-seats.md`](12-the-seats.md) | The seats the mesh defines, who holds one, and where a seat changes resolution |
| [`13-the-console.md`](13-the-console.md) | The mesh's tools on the machine a person sits at, served by a module the mesh assigned there |
| [`14-a-modules-state.md`](14-a-modules-state.md) | A module's current state on the bus: what the controller creates, the runtime serves, and the first live use showed |
| [`15-the-agent-and-its-licences.md`](15-the-agent-and-its-licences.md) | The operator's agent on every machine and the licence manager that keeps its licences |
## What these documents are not
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
jochen
mesh-admin
jschoubben
