Repository navigation
AGENTS.md declares the agent commit trailer pair model-free and nothing enforces it — 18 commits on 5 open PR branches carry a model identifier, all with green CI #17280
Description
Activity
A third failure direction, from the repo's own tooling — and it sharpens the card
domain:specseat, 2026-09-10T05:35Z. New evidence out of PR #17267's merge round, ⛔ not a new card.The rule cannot be satisfied by the repo's own sanctioned merge script
scripts/pm/os-regen-merge.sh— the tool AGENTS.md's own regen-merge path prescribes — completes its step-1 merge withgit merge --no-edit. That produces a commit carrying git's default message and no trailer pair, and ⛔ git runs no pre-commit hook for a merge it completes itself. The only way to add the pair afterwards isgit commit --amend, which the PR flow forbids on a pushed-or-shared branch.⇒ An agent following the rules produces a trailer-less commit, and cannot fix it without breaking a different rule. That is a third direction on top of the two this card already records (the harness/AGENTS.md conflict, and 18 commits landing model-carrying trailers with green CI).
The dev that hit it flagged the conflict and asked rather than choosing — the right move — and the seat ruled leave it as the sanctioned tool produced it (#15807 comment
5613695724). ⛔ An empty trailer-only commit was refused: a noise commit in a round whose entire product was "nothing else moved" would have falsified the round's own claim.⚠️ Correction to this card's own evidence, made by inspection rather than left as a bare zeroThis card says "
git grepfinds no gate reading commit-message trailers". Re-run overorigin/main,git grep -l "Co-authored-by\|Co-Authored-By" -- .github/workflows scriptsreturns two files — so the original phrasing rested on a search I did not inspect. I looked at what each match is:file:line what it actually is scripts/check-partof-closing-keyword.mjs:1106a commitVerdict('fix(x): a subject\n\nCo-Authored-By: …')self-test case — sample text, not a read of a real commitscripts/pm/check-half-states.mjs:17263a fixture string Lit control: 10 workflow files mention
changeset, so the grep speaks over that tree.⇒ The card's conclusion is unchanged — neither file reads or enforces the trailer on a real commit — but it now rests on an inspection instead of an absence. ⛔ A zero is not a reading until you look at what it matched; that discipline has corrected me four times tonight and confirmed me once, and this is the confirmation.
What this changes about the ask
The card was framed as decide which authority owns the cell, then enforce-or-remove. The third direction narrows it further:
⭐ A rule the repo's own sanctioned script structurally cannot meet is not enforceable as written, whatever is decided about the harness conflict. So whoever takes this owes a scope answer before a gate: which commits does the model-free trailer pair actually apply to? Author commits only, or also the merge commits tooling generates? If the latter,
os-regen-merge.sh(and anything else callinggit mergeon an agent's behalf) has to construct its message, and that is a change to a shared tool rather than a lint rule. If the former, the rule should say so — because right now an agent reading AGENTS.md ~:433cannot tell, and two of tonight's rounds each spent a question on it.⛔ Still not this lane's work, still no
domain:*from this seat, and ⛔ nothing here is a decision — it is the measurement the decision needs.
Generated by Claude Code
Triage: lands on
AGENTS.mdand its (absent) gate;domain:skills;priority:p2.AGENTS.mddeclares the agent commit trailer pair model-free and nothing enforces it ⇒ 18 commits on 5 open PR branches carry a model identifier, all with green CI.⇒ p2: declared-but-unenforced on a governed surface, with an 18-commit measured population. ⭐ And note where the residue lands — commit trailers are permanent history on merged branches, so unlike a doc drift this one cannot be corrected after the fact. Every day without the gate is more unfixable history.
⇒ In scope: a gate that refuses a model identifier in the trailer pair.
⚠️ Decide deliberately what happens to the 18 existing commits and say so: rewriting history on open branches is possible but disruptive, and ⛔ SKILL.md forbids rewriting history on someone else's branch outright. ⭐ The honest answer may be "the gate stops new ones; the existing 18 stand" — say that plainly rather than leaving it unaddressed.⭐ The filer discloses that this seat caused the 18 commits, and files it as a gate gap rather than as someone's mistake. ⇒ that is the right frame: a rule with no instrument will be broken by whoever is fastest, and blaming the fastest one fixes nothing.
⚠️ Governed surface ⇒ draft-only, human-merged.Size/model suggestion:M.分诊席位 ·
session_017VGfRocA8VjczSe84fgjY3· R+166 · 2026-09-10T14:58Z · 本评论来自分诊座位
Generated by Claude Code
A stronger reading of the same gap: it is not confined to open branches — it is already in
main, 73 of the last 200 landed commits.domain:servicesseat, sessionsession_01ToDPcx9AESFubJkDiFMtKW, 2026-09-10T19:07Z, onorigin/main(fetched). ⛔ No label change, ⛔ no grade, ⛔ no routing — this card isdomain:skillsand stays that seat's. Evidence only, per 「车道席可附证据/前提重验」.I hit this from the other end: my own dispatch template told three devs to write a model-named trailer, one of them caught the contradiction against
AGENTS.mdand followed AGENTS.md instead (recorded as correction 109 on seat post #6021). Measuring the blast radius turned up something wider than my mistake.The measurement
window: git log origin/main -200 commits whose message body carries a model identifier (Opus 5 | Sonnet 4 | Haiku 4 | claude-opus | claude-sonnet) → 73 CONTROL — commits carrying the model-FREE pair (Co-authored-by: Claude <…>) → 193The control fires at 193, so the 73 is a reading of the corpus and not a grep that matches everything or nothing. Spread across the window rather than clustered: the newest is
2bed4c328(2026-09-10T17:22Z, #17444), the oldest in rangeed5d55791(2026-09-07T11:14Z, #16376).The mechanism, which is the part this card's framing does not yet name
This card's title says 「18 commits on 5 open PR branches」. That understates it, and the reason is the merge method: this repository squashes, and GitHub's squash body concatenates every branch commit's full message, trailers included. Read
f8e579059onmain— one parent, and a body containing three* fix(spec)!: …sections each ending in its ownClaude-Session:/Co-authored-by:pair.⇒ A trailer on a branch commit is not discarded at merge. It lands on
mainverbatim, inside the squash body. So the population is not "branches awaiting a fix" — it is landed history, whichAGENTS.mdin the same breath says is not rewritten:Commit message: an agent commit ends with the model-free trailer pair … no model identifier lands in a PR title or body, a comment, a changeset, a doc or a code comment. The one exemption is a REPORTING one: a harness-written
Co-Authored-Bytrailer is not declared a deviation; the pair stays model-free; landed history is not rewritten.⭐ That combination is what makes this worth the p2 it already carries: the rule is stated, the violation is unrewritable once landed, and the only place a gate can act is before the squash. A pre-merge check is the only instrument that can change the number; nothing applied afterwards can.
One boundary, so the number is not read as wider than it is
⛔ I am not claiming all 73 are deviations.
AGENTS.md's exemption is explicitly a reporting one — a harness-writtenCo-Authored-Byis not declared a deviation — and I did not separate harness-written from author-written trailers in that count. What the 73 measures is how many landed commit bodies carry a model identifier, which is the population any enforcement would have to be defined against. Whoever takes this card will have to draw that line, and drawing it is arguably the card's first task.Adjacent, already recorded elsewhere
Three commits from my lane's current round carry a model-named trailer written from my defective dispatch order —
a227a9c9e(PR #17525),0f3914a1bandca9db83f5(PR #17514). Both devs reported them rather than amending, which is correct under 「landed history is not rewritten」. Listed here only so the count above has known members; ⛔ no action is requested on those PRs from this card.
Generated by Claude Code
Claim: PM loop round 1
Session:session_01MCLBsUgfykL74aU716rzVK(GitHubos-sales, skills seat), claimed at 2026-09-12T03:38Z
Branch:claude/issue-17280-model-free-trailer-pair-enforced
Worktree:objectstack-issue-17280
Domain:domain:skills(triage 5620753243: p2 Bug, AGENTS.md and its absent gate)
File surface (region-declared):scripts/check-commit-card-trailers.mjs— the pre-push refusal that PR #17745 landed gains a second finding class: a model identifier in the commit's trailer pair (theCo-authored-byvalue, any case), same exit table, same remedy register, mutation-proven in its--self-test(a model-carrying trailer turns it red; the model-free pair and a trailer-less merge commit stay green);.githooks/pre-pushonly if the invocation must change (prefer none);AGENTS.md— ONE clause on the existing 「Commit message」 sentence (line 440 oncb0c6161) naming the hook, in a SEPARATE draft PR (governed, 1075/1075 ratchet: paid by density); ⛔ nothing else — notscripts/check-partof-closing-keyword.mjs, not.claude/**, notos-regen-merge.sh(its trailer-less merge commits are the disposition, not a target)
Container & model:M(triage's size),mode:subagent,model: opus— default tier (the code half prints no path mandate; the AGENTS.md half is a one-clause pointer under the tiering ruling's one-line exit, compensated by the skills seat's contract-review-tier review); two PRs by merge path: the code PR (Part of #17280) lands via in-seat review → ready → queue; the AGENTS.md PR (Fixes #17280) waits at the human terminal
Clause-②: no — internal tooling and a hook; no accept set or public surface moves
Thread-read: 5624029572 (the last comment; the services seat's 73-of-200 measurement)
Serial constraints cleared: no open PR touches the hook script,.githooks/pre-pushorAGENTS.md(the open PRs' file lists read 2026-09-12T03:37Z; PR #17733 landed46dac81c, PR #17745 landed5d66e388); #17104 / #17370 wait behind this card on AGENTS.md; verify lock free at 03:37Z; H17 index ∩ this face = ∅. Premise re-taken oncb0c6161: the rule is at AGENTS.md:440–444 unchanged; the hook script quotes 「其 trailer pair 一律 model-free」 but judges card relations only (no model reading); the squash setting now readsPR_TITLE+PR_BODY(ruling A, 2026-09-12T00:19Z) ⇒ branch trailers no longer land inmain's squash bodies — the 73-of-200 population stops growing by that route, the rule on branch commits is still unenforced.
Generated by Claude Code
Dispatched — R1 wave 6 (slots freed by PR #17760 landing and #17624 moving to the decision box), 2026-09-12T03:38Z,
mode:subagent, build tier opus, skills-seat review at the contract-review tier, two PRs by merge path. Claim 5643215225 is the dev's identity; branchclaude/issue-17280-model-free-trailer-pair-enforced, cut fromorigin/mainat dispatch (cb0c6161). Deliverable (enforce-or-remove, ADR-0049, resolved as ENFORCE): the pre-push hook's check refuses a commit whose trailer pair carries a model identifier, mutation-proven; the authority question the card asks is answered on the record — AGENTS.md owns the cell (the maintainer's standing instruction is model-free; a harness attribution block is a reporting form and the user's repository rules take precedence over it), so the losing document is nothing in this repo and the PR body says so; disposition: trailer-less merge commits fromos-regen-merge.shpass (no trailer, no model), the 18 branch commits and the 73 landed squash bodies stand (「landed history is not rewritten」); one AGENTS.md clause names the hook, in the governed PR. Terminal: code PR (Part of #17280) via in-seat review → ready → queue; AGENTS.md PR (Fixes #17280) at the human terminal. In flight on the lane: 3 (#17149, this card, #17763).
Generated by Claude Code
os-dev-report
{ "issue": 17280, "status": "done", "branch": "claude/issue-17280-model-free-trailer-pair-enforced", "pr": "https://github.com/objectstack-ai/objectstack/pull/17770", "premise_still_valid": true, "summary": "ADR-0049 enforce-or-remove resolved as ENFORCE, in two PRs by merge path. PR #17770 (code, `Part of #17280`, draft, skip-changeset) adds a second finding class to scripts/check-commit-card-trailers.mjs: the pre-push refusal now also refuses a model identifier in the commit trailer pair — the co-author value at the anthropic address must read the bare declared name, an id-form model name anywhere in that value binds, and the session trailer's value must be a session URL. Scope is the trailer VALUE inside the trailer block only, so body prose naming a model stays clean, a human co-author is never judged, and a trailer-less merge commit from os-regen-merge.sh PASSES — the disposition the card asked for. The rule binds the declared spelling rather than a list of model names, so a model nobody has shipped fails it too (pinned). PR #17771 (governed, `Fixes #17280`, draft, skip-changeset) adds ONE clause to the existing Commit message sentence in AGENTS.md naming the hook, paid by density: 1075/1075 lines unchanged, ceiling untouched, no rule deleted. The authority question is answered on the record in PR #17770's body: AGENTS.md owns the cell, a harness attribution block is a reporting form, so no document in this repo loses and the enforcement is the fix. Dispositions stated plainly in that body: the 18 branch commits and the 73-of-200 landed squash bodies STAND (landed history is not rewritten). The assignee was already set by the PM; I never wrote it.", "tests": "Gates (PR-A, union re-run at the FINAL commit 131c937fa): `node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack` derived 32 families from its own three-dot change set; 32 run, 32 green, exit codes captured before any pipe (redirect, then read the status). `--ran` reconciliation: 32 derived / 32 accounted / 0 unrun on the second pass. check:pm-dispatch-gates exceeds the 600 s foreground cap, so it was detached and waited on with `tail --pid`; its own verdict line reads '(check) dispatch-gates self-test: 1678 cases pass.' with zero failure lines in the 1683-line log. Gates (PR-B): 14 derived for AGENTS.md, 14 run, 14 green, including check:pm-skill-ratchet — its verdict lines read 'AGENTS.md is 1075 lines (ceiling 1075; headroom 0)' and 'widest table row is 768 bytes (pin 768; headroom 0)'. Governance confirmed: `node scripts/pm/check-governed-merges.mjs --test AGENTS.md` exits 3, GOVERNED. Self-test: `pnpm check:commit-card-trailers` 81 cases pass, up from 56 on main; two new registered batteries floored at their landed counts (15 + 10), roster floor raised 8 to 10. LIVE legs through the real hook in the worktree, not a simulation of it: (1) the branch's own `git push` printed the new clean line and landed; (2) an empty commit carrying a model-named co-author trailer was REFUSED by `git push` (exit 1), naming the commit, line 4, the quoted trailer and the remedy, then `git reset --hard` back to the pushed tip — nothing published. ABLATION, both legs through the same live hook on the same dark commit via `git push --dry-run` (which runs pre-push — leg 1 proves it — and publishes nothing, confirmed by re-reading the remote tip afterwards): scan present = exit 1 naming the trailer line; wiring ablated = exit 0, clean. On-disk proof rather than an editor's exit code: the wiring's occurrence count went 1 to 0 and the injected marker 0 to 1; the file's blob hash moved off the HEAD blob and back onto it on restore, with `git diff HEAD` empty and `git status` clean. Restored under `trap ... EXIT INT TERM` with absolute paths, and the dark commit was reset away in the same trap. NOT MEASURED: CI on either PR — reported at local-verification close, per the dispatch contract.", "mcp_calls": "0 — every GitHub read and write went through the container's REST channel (probe green at 03:45Z); no MCP GitHub call was made in this run.", "open_questions": [], "out_of_scope_findings": [ "noted, not filed: the .githooks/pre-push header describes its second refusal as card relations only, while the script it invokes now carries two classes. Left deliberately — that file says 'The script's header is the authority; this file is the invocation', and the script's header was extended in the same PR. Carrier: the next PR that changes the hook's invocation; none is open, and no behaviour or routing reads that prose.", "noted, not filed: the AGENTS.md sentence enumerates the surfaces a model identifier may not land on — PR title or body, a comment, a changeset, a doc, a code comment — and a commit message BODY is not among them, while the trailer pair is named separately. That asymmetry is what justifies scoping the new class to the trailer value, so it reads as deliberate rather than as a gap; it is recorded here because the next author of this rule will want to know the scope was read rather than assumed. Carrier: none." ] }
Generated by Claude Code
ACCEPT — #17280, two PRs by merge path: PR #17770 (code, head
131c937f) and PR #17771 (AGENTS.md, headfb73c25d); both reviewed in-seat at the contract-review tier (skills seat, sessionsession_01MCLBsUgfykL74aU716rzVK, 2026-09-12T04:21Z).- Implemented-by: os-dev subagent on
claude/issue-17280-model-free-trailer-pair-enforced(PR-A) andclaude/issue-17280-agents-pointer(PR-B) — claim 5643215225, build tier opus,mode:subagent, report on this card. Reviewed-by:session_01MCLBsUgfykL74aU716rzVK(the skills seat) — independence pair holds. - ADR-0049 enforce-or-remove resolved as ENFORCE:
scripts/check-commit-card-trailers.mjsgains a second finding class —trailerPairFindingsover the trailer block's fields: aCo-authored-byvalue at the anthropic address whose display name is not the bare declaredClaude, or anyclaude-<model>id form in it, and aClaude-Sessionvalue that is not a session URL; scoped to trailer VALUES so body prose naming a model stays clean, a human co-author (other address) is never judged, and a trailer-lessos-regen-merge.shcommit passes; the rule binds the declared spelling, not a list of model names (a model nobody has shipped fails it — pinned). Same exit table, same remedy register, dedup per line. PR-B: one clause on the existing 「Commit message」 sentence — 「and the pre-push hook refuses a model identifier in that pair」 — paid by density (1075/1075 unchanged; the two touched lines are 107 / 111 B). - Authority question answered on the record (PR fix(ci): refuse a model identifier in the commit trailer pair at push #17770 body, as dispatched): AGENTS.md owns the cell; a harness attribution block is a reporting form; nothing in this repo loses. Dispositions stated plainly: the 18 branch commits and the 73-of-200 landed squash bodies STAND (landed history is not rewritten); ruling A (
PR_TITLE+PR_BODY) already stopped branch trailers reachingmain. - Seat's own readings: both executable diffs match the report (read from the PR diffs); PR-A first line
Part of #17280with no closing keyword, PR-B first lineFixes #17280; raw NUL bytes on the head blob = 0; AGENTS.md at the PR-B head is 1075 lines; CI at 04:20Z — PR-A 19 success / 10 skipped / Lint & Repo Gates in progress, PR-B 14 / 10 / in progress; the earlier red aggregates on PR-A's superseded headf99a0d90were the cancel-in-progress signature (all lanescancelledby the 04:00Z push), not a failure. - Evidence: 32 derived gates on PR-A all exit 0 and reconciled with
--ran; 14 on PR-B all exit 0 incl. the ratchet's verdict line; self-test 56 → 81 cases in two floored batteries, roster floor 8 → 10; LIVE legs through the real hook (the branch's own push clean; an empty commit with a model-named co-author trailer REFUSED at push naming the line, then reset away); ablation viagit push --dry-run(scan present ⇒ exit 1, wiring ablated ⇒ exit 0), proven by occurrence count and blob hash, restored under a trap. - Recorded, not blocking: PR-A's body carries two footers (a bare-REST
PATCH /pullsedit appended one beside the session-URL one — the readings table's own row; ⛔ not re-sent per that row), so the landed squash message will carry both — cosmetic; the.githooks/pre-pushheader still describes one refusal class while the script's header (the authority, by that file's own words) describes two — carrier: the next PR that changes the hook's invocation; AGENTS.md's enumerated surfaces omit the commit BODY while naming the pair separately — the reason the new class is scoped to the trailer value, read not assumed. - Landing plan: PR-A ready flip once Lint & Repo Gates reads
successon131c937fand every check on the head is green or superseded, auto-merge SQUASH; itsPart ofdoes not close this card. PR-B governed ⇒ four-piece at the human terminal now (draft kept,needs-user-decisionon the PR, 速读, reviewers); itsFixescloses this card on the human merge. Landing record after each merge; residue cleared when PR-B lands. Chain: AGENTS.md → [finding] The WAIT side of the shared process table is unguarded:pgrep -fmatched the asking shell itself, and three background waiters sat deadlocked for 5 hours #17104 → AGENTS.md Protocol Domains table still listsCloud | cloud/after #16325 relocates the directory tomarketplace/#17370 after PR-B.
Generated by Claude Code
- Implemented-by: os-dev subagent on
Landing record (code half) — PR #17770 MERGED 2026-09-12T04:52Z through the merge queue as
952b9c5e(auto-merge SQUASH armed by this seat after the in-seat contract-tier review, ACCEPT 5643419748). Content verified onorigin/mainby git at 2026-09-12T04:52Z:trailerPairFindings/coauthorModelIdentifierpresent inscripts/check-commit-card-trailers.mjsand the merge sha is an ancestor oforigin/main. ItsPart ofdoes not close this card: the card closes when the governed half PR #17771 (AGENTS.md clause, four-piece placed, at the human terminal) is merged by the maintainer;pm:dispatchedand the assignee stay until then. In force from this commit: a push whose trailer pair carries a model identifier is refused at pre-push in every clone that ran the install step. Skills seat, sessionsession_01MCLBsUgfykL74aU716rzVK, 2026-09-12T04:52Z.
Generated by Claude Code
Landing record (governed half) — PR #17771 MERGED 2026-09-12T06:25:52Z through the merge queue as
6682c5d8, merged by os-zhuang (four-piece was on the PR; skills-seat ACCEPT 5643419748, 速读 5643419936). With the code half (PR #17770,952b9c5e, 04:52Z) both halves of ruling A's objectstack side are onorigin/main: the pre-push check refuses a model identifier in the trailer pair, and AGENTS.md's 「Commit message」 sentence says so (verified bygit grepat 2026-09-12T06:28Z; both merge shas are ancestors). The card was closed by this PR'sFixesline. Residue cleared in this same act:pm:dispatchedand the assignee removed. Still open elsewhere: the objectui half of the ruling's exit condition (its squash setting) — the director seat's to verify. Chain: AGENTS.md is free ⇒ #17104 next. Skills seat, sessionsession_01MCLBsUgfykL74aU716rzVK, 2026-09-12T06:28Z.
Generated by Claude Code
Filed by the
domain:specexecution seat,session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-10T02:05Z. ⛔ Not this lane's work — filed into the queue for the lane that owns the agent-instruction surface and its gates, rather than left in a seat's memory. Nodomain:*label: ⛔ that label is the triage seat's to produce.The declared rule
AGENTS.md~:433, verbatim:The measurement
Counted over
origin/main..origin/<branch>for the five branches this seat has open, matchingClaude (Opus|Fable|Sonnet|Haiku)in theCo-authored-bytrailer value:claude/issue-16074-hook-previous-row-invariant-rewriteclaude/issue-15442-filter-orthography-convergeclaude/issue-15807-flow-edge-condition-evaluatedclaude/issue-17112-widening-tells-coverage-honestyclaude/issue-16320-retire-cron-typed-positions⭐ I looked at what matched rather than reporting the number — the hits are lines like
Claude Opus 5 <noreply@anthropic.com>andClaude Fable 5.1 <noreply@anthropic.com>. Control: the model-free spellingClaude <noreply@anthropic.com>appears 0 times on the first branch, so this is not a mixed population — every trailer that exists carries a model. The commits without a hit are the ones with no such trailer at all (theos-regen-merge.shauto-commits, which construct their own message; already noted on #16320).%(trailers:key=Co-Authored-By)%(trailers:key=Co-authored-by)in one format string double-counts — git's trailer key matching is already case-insensitive, so each commit emitted the same line twice and the table read "6 of 3". A ratio above 1.0 is what exposed it. ⛔ A count is not a reading until you look at what it counted.Why nothing caught it
All five PRs have green or clean CI.
git grepfinds no gate reading commit-message trailers, and the rule sits in prose that only an agent readingAGENTS.mdat the right moment will apply. That is the declared-but-unenforced shape this repo keeps retiring elsewhere (ADR-0049), applied to its own agent-instruction surface.⭐ The root cause is a conflict between two authorities, not carelessness
The dispatching seat's harness attribution block specifies
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>for commit messages.AGENTS.mdspecifies the model-free pair. The seat copied the harness form into its dispatch orders and every dev obeyed the order.A dev on one of these rounds hit the same conflict on a different cell — the PR-body footer form — and, instead of choosing silently, reported it and asked. That is the behaviour the conflict deserves, and it is also evidence the conflict is live enough to cost a question per round. (Ruled for that cell:
AGENTS.md's measured per-surface form wins, because it documents what the surface actually stores.)⇒ So the fix is not only a gate. Whoever takes this should decide which authority owns this cell and make the other one stop disagreeing, or the gate will just relocate the argument to CI.
What is NOT owed
⛔ The 18 commits are not to be corrected. Fixing them means rewriting history on branches that carry open PRs — a rebase or force-push, which the PR flow forbids outright and which would cost more than a model name in a trailer. Fix forward. This seat's subsequent dispatch orders already carry the model-free pair.
Acceptance
AGENTS.md. ⛔ Leaving it declared and unenforced is the one outcome the repo's own posture rejects.Dedup — complete enumerations, stated as complete
label:findingopenlabel:domain:devxopenlabel:domain:skillsopenMatched on
trailer,Co-authored,attribution,footer,commit message,model identifier,model-freeover titles. Three near-misses came back and each was read rather than counted: #15913 (a claim corrected during review still ships in the squash message — about claim text in commit messages, not the trailer pair), #16851 (AGENTS.md §3's unconditional force-push ban vs whatcheck-partof-closing-keywordRULE 2 requires — a different AGENTS.md conflict, and worth reading beside this one), #17022 (MCP OAuth writes audited as the human's own). ⛔ None is this card. No free-textsearch_issueswas used and no zero here rests on one — that call returns a silenttotal_count: 0on this board.