Amends to-be 55 (the knowledge base) for issue 504, after the operator's decision of 2026-10-12 noted on that issue.
What changes
§5: a whole-set rule that fails does not make the module unhealthy. It is told to the operator once, as a warning naming the rule and the records, and cleared once the rule holds again. The module's health names it and stays healthy, because health says only whether the module works (issue 508's principle), so a finding never holds a walk at the module's first-node gate.
§5: the rule cited is also judged in the review, before the judgement. A decision with no home and no accepted record linking it gets changes from the knowledge base itself. It is not refused at propose, because no record can link a number that is only reserved.
§10: the row "a whole-set rule that fails is loud" now names the warning and the healthy module. A new row covers cited judged in the review.
Decision. ADR 0290 §4 says the whole-set rules "raise a condition when one fails". It does not say the module turns unhealthy, so a warning to the operator still meets it. The operator's decision on issue 504 settles how. No new decision record is written. If the record-keeping rule wants one, that is a separate step.
Checks.MESH_CATALOG_DIR=<mesh-catalog at #237> sh merge-check.sh passes: records, index, cycle, words. Public-repo rules: no node names, addresses or paths.
By fix-504@g14. Not merged.
Amends to-be 55 (the knowledge base) for issue 504, after the operator's decision of 2026-10-12 noted on that issue.
**What changes**
- §5: a whole-set rule that fails does not make the module unhealthy. It is told to the operator once, as a warning naming the rule and the records, and cleared once the rule holds again. The module's health names it and stays healthy, because health says only whether the module works (issue 508's principle), so a finding never holds a walk at the module's first-node gate.
- §5: the rule `cited` is also judged in the review, before the judgement. A decision with no home and no accepted record linking it gets changes from the knowledge base itself. It is not refused at `propose`, because no record can link a number that is only reserved.
- §10: the row "a whole-set rule that fails is loud" now names the warning and the healthy module. A new row covers `cited` judged in the review.
**Decision.** ADR 0290 §4 says the whole-set rules "raise a condition when one fails". It does not say the module turns unhealthy, so a warning to the operator still meets it. The operator's decision on issue 504 settles how. No new decision record is written. If the record-keeping rule wants one, that is a separate step.
**Built by** mesh-catalog #237 (the same issue).
**Checks.** `MESH_CATALOG_DIR=<mesh-catalog at #237> sh merge-check.sh` passes: records, index, cycle, words. Public-repo rules: no node names, addresses or paths.
By fix-504@g14. Not merged.
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.
Deliverynovox/hq@fe4f62c712dd — delivered since 2026-10-12T11:45:28Z
Delivery plan — builds nothing: the change touches no module of the mesh's graph
Transitions
2026-10-12 11:28 (new) → proposed (announced): novox/hq#325's head announced
2026-10-12 11:31 proposed → checked (checked): the verdict names this commit
2026-10-12 11:31 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
2026-10-12 11:45 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
2026-10-12 11:45 published → delivered (done): nothing for a walk to move
The commit's note under refs/notes/mesh-plan keeps every transition: git log --notes=mesh-plan.
<!-- mesh-delivery:view -->
**Delivery** `novox/hq@fe4f62c712dd` — **delivered** since 2026-10-12T11:45:28Z
**Delivery plan** — builds nothing: the change touches no module of the mesh's graph
**Transitions**
- 2026-10-12 11:28 (new) → proposed (announced): novox/hq#325's head announced
- 2026-10-12 11:31 proposed → checked (checked): the verdict names this commit
- 2026-10-12 11:31 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
- 2026-10-12 11:45 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
- 2026-10-12 11:45 published → delivered (done): nothing for a walk to move
The commit's note under `refs/notes/mesh-plan` keeps every transition: `git log --notes=mesh-plan`.
Independent review of the to-be 55 amendment for issue 504.
Checks. mesh/repo-check passes, and mesh/merge-gate passes (it touches no module). Locally at fe4f62c, records.py, cycle.py, index.py and words.py all pass.
Public-repo rules. The diff has no node names, addresses, domains or absolute paths. It uses role and design words only ("the module's first-node gate") and cites issues 504 and 508 by number.
§5: a whole-set finding is told once as a warning, cleared when the rule holds again, and named in the module's health while the module stays healthy.
§5: cited is judged in the review before the judgement, not at propose.
§10: the two new rows say how each rule is checked, and both name tests that exist in #237 (TestADecisionNothingWouldCiteIsNotAccepted, and TestARecordAcceptedCitingANewDecision... with its no-repeat-after-restart and clear steps).
Does the change need a decision record? I think not. ADR 0290 §4 says the whole-set rules "run over it after every acceptance and raise a condition when one fails". It does not say that the module's health turns unhealthy. The glossary defines a condition as "one open fact about something the mesh owns that is wrong, with a key and a severity; raised and cleared only by observation". The new notice fits that definition: it has a key (knowledge.finding.<rule>.<records>) and a severity (warning), it is raised when the keeper observes the failure, and it is cleared when the rule is observed to hold. The messenger routes it "like the controller's conditions". So the operator's decision settles how §4 is met, not what §4 decided. Issue 508 set the same precedent for a proposal that was never reviewed.
There is one honest gap in that reading. The fact is no longer among the controller's conditions (mesh-controller.conditions and doctor will not list it). It reaches the operator through the operator's channel instead. If that ever matters, a progressive insight on ADR 0290 §4 that says how the condition is raised would make it exact. It is not needed for this pull request.
Not blocking. One line in §5 runs past the file's wrap width ("…first-node gate (the operator's decision of 2026-10-12 on issue 504, as issue 508 decided for a proposal never"). This is cosmetic.
Verdict: approve at fe4f62c
Independent review of the to-be 55 amendment for issue 504.
**Checks.** mesh/repo-check passes, and mesh/merge-gate passes (it touches no module). Locally at fe4f62c, records.py, cycle.py, index.py and words.py all pass.
**Public-repo rules.** The diff has no node names, addresses, domains or absolute paths. It uses role and design words only ("the module's first-node gate") and cites issues 504 and 508 by number.
**It matches what mesh-catalog #237 builds.**
- §5: a whole-set finding is told once as a warning, cleared when the rule holds again, and named in the module's health while the module stays healthy.
- §5: `cited` is judged in the review before the judgement, not at `propose`.
- §10: the two new rows say how each rule is checked, and both name tests that exist in #237 (`TestADecisionNothingWouldCiteIsNotAccepted`, and `TestARecordAcceptedCitingANewDecision...` with its no-repeat-after-restart and clear steps).
**Does the change need a decision record? I think not.** ADR 0290 §4 says the whole-set rules "run over it after every acceptance and raise a condition when one fails". It does not say that the module's health turns unhealthy. The glossary defines a condition as "one open fact about something the mesh owns that is wrong, with a key and a severity; raised and cleared only by observation". The new notice fits that definition: it has a key (`knowledge.finding.<rule>.<records>`) and a severity (warning), it is raised when the keeper observes the failure, and it is cleared when the rule is observed to hold. The messenger routes it "like the controller's conditions". So the operator's decision settles how §4 is met, not what §4 decided. Issue 508 set the same precedent for a proposal that was never reviewed.
There is one honest gap in that reading. The fact is no longer among the controller's conditions (`mesh-controller.conditions` and `doctor` will not list it). It reaches the operator through the operator's channel instead. If that ever matters, a progressive insight on ADR 0290 §4 that says how the condition is raised would make it exact. It is not needed for this pull request.
**Not blocking.** One line in §5 runs past the file's wrap width ("…first-node gate (the operator's decision of 2026-10-12 on issue 504, as issue 508 decided for a proposal never"). This is cosmetic.
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.
Amends to-be 55 (the knowledge base) for issue 504, after the operator's decision of 2026-10-12 noted on that issue.
What changes
citedis also judged in the review, before the judgement. A decision with no home and no accepted record linking it gets changes from the knowledge base itself. It is not refused atpropose, because no record can link a number that is only reserved.citedjudged in the review.Decision. ADR 0290 §4 says the whole-set rules "raise a condition when one fails". It does not say the module turns unhealthy, so a warning to the operator still meets it. The operator's decision on issue 504 settles how. No new decision record is written. If the record-keeping rule wants one, that is a separate step.
Built by mesh-catalog #237 (the same issue).
Checks.
MESH_CATALOG_DIR=<mesh-catalog at #237> sh merge-check.shpasses: records, index, cycle, words. Public-repo rules: no node names, addresses or paths.By fix-504@g14. Not merged.
Delivery
novox/hq@fe4f62c712dd— delivered since 2026-10-12T11:45:28ZDelivery plan — builds nothing: the change touches no module of the mesh's graph
Transitions
The commit's note under
refs/notes/mesh-plankeeps every transition:git log --notes=mesh-plan.Verdict: approve at
fe4f62cIndependent review of the to-be 55 amendment for issue 504.
Checks. mesh/repo-check passes, and mesh/merge-gate passes (it touches no module). Locally at
fe4f62c, records.py, cycle.py, index.py and words.py all pass.Public-repo rules. The diff has no node names, addresses, domains or absolute paths. It uses role and design words only ("the module's first-node gate") and cites issues 504 and 508 by number.
It matches what mesh-catalog #237 builds.
citedis judged in the review before the judgement, not atpropose.TestADecisionNothingWouldCiteIsNotAccepted, andTestARecordAcceptedCitingANewDecision...with its no-repeat-after-restart and clear steps).Does the change need a decision record? I think not. ADR 0290 §4 says the whole-set rules "run over it after every acceptance and raise a condition when one fails". It does not say that the module's health turns unhealthy. The glossary defines a condition as "one open fact about something the mesh owns that is wrong, with a key and a severity; raised and cleared only by observation". The new notice fits that definition: it has a key (
knowledge.finding.<rule>.<records>) and a severity (warning), it is raised when the keeper observes the failure, and it is cleared when the rule is observed to hold. The messenger routes it "like the controller's conditions". So the operator's decision settles how §4 is met, not what §4 decided. Issue 508 set the same precedent for a proposal that was never reviewed.There is one honest gap in that reading. The fact is no longer among the controller's conditions (
mesh-controller.conditionsanddoctorwill not list it). It reaches the operator through the operator's channel instead. If that ever matters, a progressive insight on ADR 0290 §4 that says how the condition is raised would make it exact. It is not needed for this pull request.Not blocking. One line in §5 runs past the file's wrap width ("…first-node gate (the operator's decision of 2026-10-12 on issue 504, as issue 508 decided for a proposal never"). This is cosmetic.