From the delivery-time audit of 2026-10-16: seven review rounds on issue 704, a feature branch's first
check failing on 29 problems, mesh-catalog #300 waiting 12 h unmarked, the mail modules failing live,
98 issues held in fixing.
The research playbook named acceptance tests the milestones carry, which
playbooks 02 and 04 did not know; its pre-check also disagreed with the
knowledge base's deferral rule.
Proposing records one after another serialises their reviews and the
operator's asks; retyping the effort's findings into each record makes them
long and gives the review more to send back.
Research 053's identity part took six versions and about two hours: options
that broke the mesh's own rules were offered and argued back, a version was
proposed after every answer, and reviews returned findings a pre-check
would have caught. A lab run was planned where an in-process test answered.
The operator found 04 odd on 2026-10-15: it went from a settled design straight to building,
with no work breakdown, and contradicted playbook 07 on pull requests. 01 now starts from the
operator's idea with setup questions, findings and a direction the operator confirms; 02
graduates that direction into one record per reversible choice and hands the design on; 04
breaks it into milestones usable on their own, one issue per repository part, each milestone
one feature step of 07. 01 notes the knowledge base's insight-only rule for an active effort
until issue 844 is delivered.
A feature waited a day for its last issue, each small issue paid for its context again, and
waits nobody saw cost the most time. The operator set the flow on 2026-10-15: prepared context
per repository, one to two hours of local development on unit tests, one review of at most three
rounds back to the same agents, one push for the integration checks, then the delivery plan;
steps for a larger feature; a separate playbook for small fixes; models per step; a wait list
with bounds; a prepared clear at each phase end.
A temporary-directory quota failure, a delivery group's manifest-check
ordering, and an agent that stops or finishes without committing each cost
time during that feature because the playbook did not yet say what to do.
Writing them down so the next feature avoids the same detours.
On 2026-10-15 issues spanning one repository finished in two to four minutes, while those spanning two or
three ran for thirty to sixty, mostly merging the feature head and checking again as other agents landed.
The operator asked for agents that finish in five to ten minutes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
They lived in a file on one machine that only one session's agents read. They are how every feature's
agents work, so they belong with the method; a feature keeps only its local facts (paths, branch, locks).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The operator, 2026-10-15: the smaller model is as good as free against the weekly allowance, so the
limit that matters is how many agents the lead can settle and follow (three to five); subagents on the
larger model stay at two at most and as rare as possible.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A reviewer that does not know its round keeps finding the next smaller thing: one decision record went
through nine rounds on 2026-10-15, each spending the operator's shared weekly allowance. The operator
asked for three to five rounds, the round told to the reviewer, and the allowance in every report.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Five agents at once, each stopping on an undecided design question and restarting, spent a
week's allowance in a night. Also write down the lessons of feature 57: never land on a red
check, a new manifest word lands with the controller that reads it, and review the tree that
is pushed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A push starts the forge's checks, which build the modules and run every
test for up to a quarter of an hour, so a review finding fixed after the
push costs another full run (the operator, 2026-10-14).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A finished issue's agent is gone, so left in fixing it goes stale; the lead
takes it over until the feature is delivered. One repository is one
delivery, not a group; readiness comes from depends-on links (review
comment 12466).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Feature 57 was first worked as a pull request per issue into protected
branches and lost most of a day waiting on merge checks. Record the issue
as the unit, a review and checks once at the end, landing by fast-forward
under a lock, and the lessons met on the way.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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.
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.
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.
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.
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).
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 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.
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.
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.
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.
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
jschoubben
mesh-tools
jochen
mesh-adminandClaude Opus 5.5