Conversation
Blog was stood up on 2026-08-01 and cut release `1.0.11` the same day, but it was never added to `registry/repos.json`. Every hub tool is registry-driven, so `spec/audit.py` and `spec/fidelity_honesty.py` had never measured it, and `reports/divergences.md` was under-reporting the fleet by exactly one repo while reading as complete. This catalogs it and lands its first audit. ## The entry records what ran, not what was predicted `reports/conformance-matrix.md` already named Blog as the reference repo for the `source-only` + `release` shape, and predicted `releaseTrigger: none` with an empty `publish[]` until a deploy ran. What actually shipped is a dispatch-only publisher that cut a tag and a source archive, so the entry is `dispatch-only` with the GitHub release declared, and the matrix row is corrected to match. The VPS deploy still has not run, so the static-site type stays deferred to #456 and `publish[]` gets revisited when it does. ## What the audit found Nine findings, no defects, verdict **operational**. Every one is the hub advancing after the carry rather than the repo regressing: - **Eight verbatim units behind the canonical.** Six are stale copies. Two never arrived at all, `AGENTS.md > Fleet Bootstrap` and `GOVERNANCE.md > Representative Data in Agent-Authored Text`, and that distinction matters because a stale copy still states the rule in an older form while an absent one states nothing. - **One undeclared section**, `AGENTS.md > Project Conventions`, whose four rules are genuinely repo-specific and duplicate no verbatim section. The convergence PRs against Blog follow separately, one per drift class per AUDIT.md section 10. ## Measured against `main`, with `develop` recorded separately The run reads hub `main` (`3a7cc64`) per AUDIT.md section 1. A re-run from a tree at `develop` (`362aec8`) reports two more re-vendors and two DEFECTs, all four from #545 taking `bypass_actors` out of the ruleset payloads an hour earlier. Those are recorded in the report rather than counted, because measuring against un-promoted hub content reports work in flight as a conformance failure. The same four will appear against every fleet repo when that promotes, and the live-ruleset half is a settings change on a protected branch, so it is yours to apply rather than an agent's. ## Two spec questions, raised rather than resolved 1. **The hub contradicts itself on whether a repo may carry its own `AGENTS.md` section.** `AGENTS.md` says a project's own conventions live in that project's `AGENTS.md`. `spec/section-model.md` says an undeclared section is drift to reconcile and lists four destinations, none of them `AGENTS.md`. A repo following the one is flagged for violating the other, and Blog is exactly that case. 2. **The character-set tiers do not classify the COPYRIGHT SIGN**, which any repo generating a feed or a rendered document will hit. A third, smaller one: the report directory here is `reports/blog/`, lowercase, matching every existing report directory rather than the repo's CamelCase name. AUDIT.md section 8 writes it as `reports/<repo>/audit.md`, which reads as exact case, and the two have quietly disagreed since the first report. Worth settling in the wording, and worth settling before someone clones on a case-insensitive filesystem. ## Verification - `python3 spec/validate.py`: 22 cataloged, 0 backlog, classifying cleanly. - `python3 spec/audit.py Blog` from a tree at hub `main`: 9 drift, 0 defect/letter/error. - `python3 spec/fidelity_honesty.py --report`: regenerated, and the whole diff is Blog appearing in the rows it belongs in. - `python3 scripts/prose_lint.py reports/blog/audit.md`: clean. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
…551) Regression fix. #545 removed `bypass_actors` from the ruleset payloads, and **every repo in the fleet has been reporting two ruleset DEFECTs since it promoted**. ## What broke `spec/audit.py` compares the live ruleset against the payload over a fixed subset, and that subset included `bypass_actors`. A payload that deliberately declares no bypass list, against a live ruleset that still has one, is a normalized diff — so the audit called it divergence: ```text == Utilities (csharp, nuget; release) @ main@e179288 == DEFECT ruleset: develop diverges from repo-config/develop.json (normalized diff) DEFECT ruleset: main diverges from repo-config/main.json (normalized diff) ``` Reproduced on Utilities, PlexCleaner and Blog. It is every repo, because every live ruleset carries the entry the payloads stopped declaring. ## The fix, and why it is this way round The field is dropped from the compared subset rather than restored to the payloads. Who may bypass a ruleset is a per-repository human decision taken in the UI, and `configure.sh` already treats it that way after #545: `apply` writes the live list back unchanged, `check` reports it without asserting. The audit asserting the same field contradicted that directly. **The defect was two tools comparing one field under opposite policies, not the field's value.** The comment says so and records what happened when it was not, because the next person to look at that list will otherwise put it back. ## How it was found While reviewing the Blog registration in #547. That report claims nine drift and no defects, which was true when it was written against hub `main` at `3a7cc64`, and stopped being true an hour later when #545 promoted. The PR body had even predicted it: "The same four will appear against every fleet repo when that promotes." It did. This is the same staleness the audit's own run-stamp rule exists to expose, arriving from the hub's side rather than the repo's. ## Verification - Utilities: 2 ruleset DEFECTs before, **0** after. Its remaining two findings are LETTER results for absent `GOVERNANCE.md` and `OPERATIONS.md`, which are the real propagation gap rather than this regression. - PlexCleaner: same, 2 before and 0 after. - `spec/validate.py` OK, diff-scoped `prose_lint` clean, editorconfig clean, syntax checked. ## Follow-on #547's `reports/blog/audit.md` should be regenerated against this corrected tool before it merges, so the committed report matches what the audit now reports rather than what it reported an hour before the payload change landed. --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Closes #549, raised by the Blog agent from a Copilot review on a downstream re-vendor and declined there because `configure.sh` is carried `verbatim`. ## The defect The payload-driven ruleset comparison added in #540 built its normalizer on `walk/1`, which arrived in **jq 1.6**. On a host carrying jq 1.5 the filter does not degrade — it **fails to compile**, so `check_ruleset` reports drift on every parameterized rule it never actually compared. That is the inverse of the false clean #540 was written to close, and arguably worse: a false *failure* teaches an operator to distrust the tool, where a false pass merely fails to warn them. ## Verified on a real jq 1.5, not argued ```console jq version: jq-1.5-1-a5b5cbe --- OLD filter (calls walk): def n: walk(if type=="array" then sort else . end); n jq: 1 compile error --- NEW filter (defines its own): {"allowed_merge_methods":["merge","squash"],"nested":{"deep":[1,2,3]},"required_status_checks":[{"context":"a"},{"context":"b"}]} ``` The reported failure reproduces exactly, and the fix compiles and returns the sorted document. Output is byte-identical to jq 1.7 on the same input, including a deliberately nested array to confirm the local definition still recurses. Finding a genuine jq 1.5 took three attempts — `imega/jq:1.5` does not exist, Debian buster's archives are gone, and Alpine 3.8 ships a master build that already has `walk`. Ubuntu 18.04 has it. Worth recording, because "I could not reproduce it" would have been the wrong conclusion from the first two. ## The choice `walk` is defined inside the filter, so the script calls nothing jq 1.5 lacks and keeps one code path across versions. The alternative in the issue — assert a jq version up front and fail with a message — was considered and declined for the reason the issue itself gives: `check` is the read-only mode, and refusing to run at all is a worse outcome than running. The issue stated that as a preference rather than a decision, and I agree with it. ## Verification `bash -n` clean, diff-scoped `prose_lint` clean, and a live `check` against this repo still passes on all three parameterized rules of both rulesets. ## Downstream `configure.sh` is carried `verbatim` with `appliesTo: "*"`, so this joins the re-vendor the ledger already lists. It is the same file that generated a re-vendor three times today, which is precisely the evidence behind the vendored-tooling entry added in #546. --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Pull request overview
Promotes develop to main while repairing two fleet-wide auditing/comparison regressions (ruleset bypass list handling in spec/audit.py, and jq 1.5 compatibility in repo-config/configure.sh), and catalogs the Blog repository with its first recorded audit/report updates.
Changes:
- Stop
spec/audit.pyfrom assertingbypass_actorsin ruleset comparisons, aligning it withrepo-config/configure.shbehavior. - Make
repo-config/configure.shruleset normalization portable to jq 1.5 by defining a localwalk-equivalent (w/1) instead of callingwalk/1. - Catalog
Bloginregistry/repos.jsonand update generated reports/matrix to include the new repo and its audit status.
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| spec/audit.py | Removes bypass_actors from the ruleset comparison subset and drops related normalization logic. |
| repo-config/configure.sh | Replaces walk/1 usage with a locally-defined recursive walker to support jq 1.5 in check mode. |
| registry/repos.json | Adds the Blog repository entry to the fleet registry. |
| reports/blog/audit.md | Adds Blog’s first audit report. |
| reports/divergences.md | Updates the generated divergence report to include Blog in relevant rows. |
| reports/conformance-matrix.md | Updates the shape reference row for Blog with the audit date and corrected notes. |
Both findings from the review on promotion #555. A promotion's head is `develop`, so the fix lands here and #555 picks it up. ## The report predicted, and the prediction was wrong `reports/blog/audit.md` closed a paragraph with: > ...and the same four will appear against every fleet repo the moment that change promotes: Two things wrong with that, and the second is why it matters. **A report records what a run observed.** A prediction cannot be verified by re-running the audit, which is the one check the run-stamp discipline offers a reader. **And it turned out to be false.** The change promoted. The four findings were then traced to `spec/audit.py` asserting a `bypass_actors` list that `configure.sh` had deliberately stopped managing, and #551 removed that assertion. So they do not appear against every fleet repo. Verified: `spec/audit.py Blog` reports **0 defect/letter/error** on this branch. A prediction goes stale the moment reality diverges and nothing re-checks it. That is the same failure the registry's `driftNote` rules already name, arriving in a report instead — and it is the second instance today of a document asserting a future that stopped being true. The sentence now ends at the reason those findings were not counted, and hands off to the enumeration that follows it. ## The comment attached its clause to the wrong noun ```diff -# Phrases in a registry driftNote that assert work is still outstanding. +# Phrases that assert work is still outstanding, matched against a registry driftNote. ``` The subject is the plural *phrases*, so `assert` was correct, but `driftNote that assert` sits adjacent and reads as a disagreement. Reordered so the clause sits with what it describes. This is the second pass over these two lines today, the first having added a missing verb. ## Verification Diff-scoped `prose_lint` clean, editorconfig clean, CRLF preserved on the report, `spec/audit.py` parses, and the audit still reports Blog at zero findings. Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 6 out of 6 changed files in this pull request and generated no new comments.
Suppressed comments (1)
repo-config/configure.sh:247
- The jq normalizer is intended to work on jq 1.5, but it uses
keys_unsorted, which is not available until jq 1.6. On jq 1.5 this can still fail to compile (similar to the original walk/1 issue), preventingcheck_rulesetfrom comparing parameterized rule parameters.
| if type == "object" then reduce keys_unsorted[] as $k ({}; . + { ($k): ($in[$k] | w(f)) }) | f
Owner
Author
|
Answering the suppressed finding on
Declined: the premise is wrong, and it was checked rather than reasoned about. jq version: jq-1.5-1-a5b5cbe
$ echo '{"b":1,"a":2}' | jq -c 'keys_unsorted'
["b","a"]
$ echo '{"p":{"m":["squash","merge"],"c":[{"context":"b"},{"context":"a"}]}}' | jq -S -c '<the shipped filter>'
{"p":{"c":[{"context":"a"},{"context":"b"}],"m":["merge","squash"]}}The builtin compiles, and the full normalizer runs on that version and returns the sorted document, including the nested array that proves the local The finding is a good shape of question, and the analogy to Worth noting for the record that this class is anticipated in the review runbook: Copilot is sometimes factually wrong about a tool's capabilities, so a finding is verified before it is fixed and before it is dismissed. This one was verified, and it does not hold. No change to this promotion. |
ptr727
added a commit
that referenced
this pull request
Aug 6, 2026
…574) Promotes the nine commits `develop` has carried since the previous promotion (#555) to `main`. Merge commit only, no squash, and `develop` is not deleted. ## What lands - **#560** Standardize the static-site-deploy repo type and its destination, including the deploy-verification guarantee and a retention rule that records which side owns the prune. - **#561** Ask for a blocked decision instead of reporting it, and scope the clickable-link rule to a surface that renders markdown. - **#564** Read a suppressed section nested inside the review-details wrapper, so `scripts/pr_review.py` stops reporting a clean round over findings that reach no thread. - **#566** Settle Markdown capitalization in `CODESTYLE.md`, and fix the two `TODO.md` prose defects the digest defect had hidden. - **#568** Check a `driftNote` on every run rather than only on an otherwise clean audit, and state what the audit does not evaluate. - **#569** Name who trips the production ref gate, and state that a harness refusal is a different thing from the maintainer's permission and is not lifted by it. - **#571** Re-audit `ptr727/Blog` against the hugo type and bump the conformance matrix. - **#573** Scope the semicolon exemption to the sentence its list lives in, with the 44 newly reported occurrences recorded on #519 rather than swept here. - **#572** Exempt a verbatim section from the coordination-reference scan. ## Issues closed Fixes #562. Closes #565. The closing keywords sit on this promotion rather than on the feature pull requests, because GitHub fires them only on a merge into the default branch. #456, #558, #563, #567, and #570 are already closed against their merged fixes. ## Verification `git merge-tree` reports no conflict between `origin/main` and `origin/develop`, so this promotion needs no throwaway resolution branch. Every constituent pull request merged green with its review loop closed.
ptr727
added a commit
that referenced
this pull request
Aug 7, 2026
Clears the `A Home for a Disproved Finding` cluster from `TODO.md`. Its `Checked` anchor was re-verified against `develop` at `756a53e` before anything was written: the file still says the reviewer is sometimes factually wrong, still requires a decline to carry evidence, and still keeps its list of known non-working request paths, with nowhere to put the proof itself. ## The gap A decline that carries proof proves something about this tree, and [GOVERNANCE.md "Every Finding Ends in an Action"](https://github.com/ptr727/ProjectTemplate/blob/develop/GOVERNANCE.md#every-finding-ends-in-an-action) is right that the thread is where it belongs while the pull request is open. Afterwards the thread is the wrong place. The pull request merges, the next round starts with no memory of the last, and the second occurrence of the same claim reaches a maintainer with no way to tell it from a first. `Disproved Claims` is a new `###` under the runbook, so the three declared `##` sections in `spec/files.json` are unchanged. An entry names the claim, what was run or read to disprove it, the revision it was proved against, and what ends it. **It is deliberately not a list to append to**: an entry outliving the code it was proved against becomes a reason not to check, which is strictly worse than proving the claim again, so an entry whose subject moves is deleted by the change that moves it rather than edited to look current. Two guards ride with it. The record answers a repeated claim and never dismisses a new one, so a finding is judged on its merits first and matched second, and a reply carries the proof re-read rather than a pointer to a file the reviewer cannot open. And the entries are this repository's own, so a repository holding a copy carries the shape and deletes an entry whose subject it does not carry. ## The three worked examples **`keys_unsorted` requires jq 1.6.** A suppressed finding on #555 against the normalizer in `repo-config/configure.sh`, reasoning by analogy from the `walk/1` failure #553 fixed. Re-run in this session rather than quoted from the thread, on `jq-1.5-1-a5b5cbe` in `ubuntu:18.04`, the build that reproduces `walk/1`: ```console jq-1.5-1-a5b5cbe $ echo '{"b":1,"a":2}' | jq -c 'keys_unsorted' ["b","a"] $ echo '{"b":1}' | jq -c 'walk(.)' jq: error: walk/1 is not defined at <top-level>, line 1: jq: 1 compile error ``` **The write-guard's fallback parse, from #601.** Declined on the ground that the arm cannot execute, since `punctuation_chars` arrived in Python 3.6, the module uses f-strings throughout, and `install.py` refuses below 3.7. That is exactly the kind of disproof that expires, which is why the entry names the floor as what ends it. It also records that the finding earned a test case rather than a change, since only `ValueError` from unbalanced quoting reaches that path in practice and nothing covered it. **The bare-SHA design, from #602.** This one came from this repository's own backlog rather than from a reviewer, and it is here because a rejected method costs the same to re-propose as a declined finding costs to re-derive, while a backlog has a place for a claim the tree contradicts and none for a method a measurement rejects. Over the 25 most recent merged pull requests the bare-SHA arm raised four references and all four were correct prose, and a path arm flagged 54 of 215 backticked candidates. Both #601 and #602 are folded under disposition **Amends "Record what was tested and against which revision, and delete an entry whose subject changes."** They land as the second and third worked examples in the shipped section rather than as `Settled` bullets, because the entry ships in the same change. ## Verification - `python3 scripts/prose_lint.py . --diff develop` clean, `markdownlint-cli2` clean on both files. - `python3 -m unittest discover -s scripts` reports 372 tests OK and `python3 spec/audit.py --selftest` passes, neither touched by this change but both run because the file is carried. - Every factual claim in the new section was read against the tree at `756a53e`: the `keys_unsorted` call in `repo-config/configure.sh`, `_git_subcommand_arglists` and `_PUNCTUATION_CHARS` in `gh-write-guard.py`, the `(3, 7)` floor in `install.py`, and the verb alternation in `pr_review.py`. ## TODO.md The cluster is deleted per the file's own step 9. What this change does not carry becomes a new `ready` cluster: `GOVERNANCE.md` outcome 2 ends at the thread and nothing agent-agnostic points at the record, so an agent that never opens the provider runbook posts a decline and has nowhere to put the proof. That is a byte-locked carried section and a fleet re-vendor, which is why it is a separate change rather than a second surface in this one. ## Downstream `.github/copilot-instructions.md` is carried `whole` at `intent` with `appliesTo: "*"`, so this joins the re-vendor the ledger already lists. A repository taking it carries the section and its rules, and starts its own entries empty. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three commits, and the first two are repairs to defects this hub shipped earlier today. Merged as a merge commit, never squashed.
71b769182e3658fe01060walkin the filter rather than calling it, for jq 1.5Why this one should not wait
mainis currently wrong about every repository in the fleet. The previous promotion tookbypass_actorsout of the ruleset payloads, butspec/audit.pycompares live-against-payload over a subset that still included that field. A payload deliberately declaring no bypass, against a live ruleset that has one, reads as a normalized diff, so every repo audits as two ruleset DEFECTs:Anyone auditing against
maintoday gets a false failure on every repository. #551 removes the field from the compared subset, because the defect was two tools comparing one field under opposite policies, not the field's value: after the bypass change,applywrites the live list back unchanged andcheckreports it without asserting, so the audit asserting it contradicted the config directly.Verified on
develop: Utilities and PlexCleaner went from two ruleset DEFECTs each to none, and Utilities' remaining two findings are LETTER results for absentGOVERNANCE.mdandOPERATIONS.md, which is the real propagation gap.The second repair
#553 closes #549. The payload-driven comparison was built on
walk/1, which arrived in jq 1.6. On jq 1.5 the filter does not degrade, it fails to compile, socheck_rulesetwould report drift on every parameterized rule it never compared. Reproduced on a genuinejq-1.5-1-a5b5cbeand confirmed fixed, with output byte-identical to jq 1.7.Both defects were found by review rather than by a gate: #551 while reviewing the Blog registration, #553 by a Copilot review on a downstream re-vendor of the same canonical.
Blog
#547 catalogs Blog, taking the registry to 22 repositories. Its committed report claims zero defects, which was true when written, briefly false while the audit regression stood, and is true again on this branch. Confirmed rather than assumed:
spec/audit.py Blogreports 0 defect/letter/error here.Verification
spec/validate.pyOK at 22 cataloged, diff-scopedprose_lintclean, editorconfig clean, live read-onlyconfigure.sh checkagainst this repo passes, and the audit reports the corrected verdicts above.