docs(adr): ADR-0131 Status is Accepted — the merge that landed it on main is the acceptance act

[os-musk]

Part of #15453

Ruling A on that card (maintainer decision batch #40, verbatim reply 「同意」, recorded by the director seat in comment 5548576472) — the second of the three records it covers. ADR-0130 landed in #15704. ADR-0128 stays with the maintainer: its landing PR #12839 was merged by the seat account os-sales, so "merge = acceptance" does not hold for that record and the ruling says in as many words that it is not covered (速读 5552091322, A/B/C unanswered).

What changed

One field, one file — docs/adr/0131-total-organization-ownership-no-null-organization-id.md. 1 insertion, 3 deletions.

Before (source lines 3–5, a single soft-wrapped field):

**Status**: Proposed (2026-09-04) — awaiting the maintainer's hand-merge, which is itself the
acceptance act for a governed surface (Prime Directive #14). ⛔ Nothing below is settled until
this record merges; the implementation cards are cut **from** the merged ADR, never ahead of it.

After (one source line):

**Status**: Accepted (2026-09-04) — accepted by the merge that landed it on `main` ([#14976](https://github.com/objectstack-ai/objectstack/pull/14976), commit `0ed271574`), which is itself the acceptance act for a governed surface (Prime Directive #14). Execution is paused under #15193 and the maintainer's standing instruction; that pause is carried there, not by this status line.

The spelling is read, not recalled: the first sentence is copied from ADR-0125 line 3 and from the ADR-0130 line this flight's predecessor landed, PR link and short commit sha included. The second sentence is the ruling's own clause, carried byte-verbatim.

The ruling's precondition, verified in this run rather than inherited

The ruling requires merged_by on the landing PR to be a human approver before any edit, and orders a stop-and-report otherwise.

• GET /repos/objectstack-ai/objectstack/pulls/14976 → merged_by: os-zhuang, merged: true, merged_at: 2026-09-04T05:37:45Z. A human approver ⇒ this record is covered.
• The date was read on two channels and they agree. git: git log origin/main --diff-filter=A names 0ed27157468d2463b48f58c2c7be634036b18c4d, author and committer date 2026-09-04T04:50:16+00:00, subject docs(adr): ADR-0131 … (#14976) — the commit that added the file, not a later migration. API: the same commit reports the same timestamp. ⇒ 2026-09-04, which is also the date the superseded Proposed line already carried.
• 0ed271574 is a real prefix of that sha (0ed27157 + 4), so the short sha the ruling names resolves.

Two reference facts checked for the reviewers rather than assumed:

• [gate] the v18 development line is not open — ADR-0131 execution cards are blocked on this card #15193 is open (read this run) — so the clause points at a live gate, which is the whole reason the ruling attached it.
• The file's only other commit since it landed is d1c86a74 (docs(adr): qualify the pre-repo ADR-0081 citations as cloud ADR-0081, letter-checked per site #15612, merged today), which qualified the pre-repo ADR-0081 citations as cloud ADR-0081 and did not touch the Status field. That PR was this file's serial holder; it is merged, and no open PR touches docs/adr/0131-*.

Two judgment calls, named because a reviewer should be able to reject them

1. The whole soft-wrapped field was replaced, not only its first source line

The ruling says "the Status line". In this record the Status field wraps over three source lines, and its tail — "⛔ Nothing below is settled until this record merges; the implementation cards are cut from the merged ADR, never ahead of it" — is the sentence that made the old field self-refuting: the record has merged, so the sentence is false on its face, and it is exactly the reading the ruled #15193 clause replaces. Leaving it would have produced a status that says Accepted and, one line later, that nothing is settled until it merges. The same call was made and accepted on ADR-0130 (#15704).

Along the four axes: ① long-term — one field that states the state and the act and nothing else is the shape ADR-0125 established and 84+ records follow; a status carrying leftover narrative is how this card's defect got written in the first place. ② real pull — the pull is measured and already happened: a domain:services seat reads this exact line as its standing pause condition. ③ AI-error prevention — the tail is the half that induces the "wait forever" failure; keeping it under an Accepted header would leave a contradiction that invites the other failure, a seat "correcting" a governed surface unasked. ④ startup focus — zero new concepts; strictly fewer words.

⚠️ If the reviewers read the ruling narrowly as "line 3 only", the fix is to restore the two trailing lines — the rest of this diff stands unchanged.

2. The ruled clause keeps a bare #15193, not a markdown link

The clause is the ruling's own text and the ruling calls it verbatim, so it is reproduced byte-for-byte. Bare #NNNN is in-corpus style for this file (dozens of sites; 16 linked ones also exist), so this costs nothing in consistency. The PR reference in the first sentence is linked, because that half is ADR-0125's template rather than ruling text.

Gates

Derived from the real change set, never a hand list — node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack at 563ba6df (stderr: "change set derived from git — 1 path(s) vs merge base ffca0df", three-dot) ⇒ 17 commands, all 17 run, reconciled: ✓ dispatch-gates --ran: 17 derived famil(ies) accounted for — 17 run, 0 NOT-MEASURED. Exit codes captured by redirect-then-capture, before any pipe.

The three the ruling names, quoted from the gates themselves:

• ✅ check-adr-links: 679 relative link destination(s) under docs/adr/ resolve
• check-adr-anchors: OK (53 anchored file(s), every governing ADR still referenced; 133 decision number(s) …; 33725 citation(s) across 4272 file(s) resolve …)
• ✅ check-adr-symbol-anchors: 2056 anchors across 139 records resolve — 270 symbol, 1761 file-level, 25 cross-repo, 6 exempt, 3 continuation. 0 line anchors survive.

Others, also quoted from their own output: check-nul-bytes: OK (scanned 8166 text file(s) … no raw ASCII control bytes)., ✓ doc authoring guard: 15062 customer-facing string(s) across 865 spec sources clean, check:cross-package-test-inputs, check:driver-memory-census, check:refd-timer-probe, check:watch-hint-literal, check:pm-governed-merges, check-closing-keyword-parity, check-comment-mask-corpus, check-ci-filter-parity, check:doc-formula-expressions.

One NOT MEASURED, stated as such rather than hidden: pnpm --filter @objectstack/lint run check:doc-formula-expressions first answered exit 3 — PREREQUISITE NOT MET (@objectstack/formula and @objectstack/lint unbuilt). After pnpm exec turbo run build --filter=@objectstack/formula --filter=@objectstack/lint under scripts/pm/os-verify-lock.sh (VERDICT command-exit 0 · held the lock 180s · waited 0s) it was re-run and exited 0. The table above reports the re-run, not the prerequisite failure.

Governed predicate, as the ruling requires: node scripts/pm/check-governed-merges.mjs --test docs/adr/0131-…md → exit 3, ⛔ GOVERNED — a human merge is the review record for this PR (#9495 regime).

pnpm lint does not apply to this diff — and that is a measurement, not a skip. Three readings, all taken at the final commit: (a) the checked population read from eslint's own configuration — eslint.config.mjs carries zero .md/markdown entries; (b) eslint --no-inline-config --format json on the one changed file returns 1 result, errorCount: 0, its single warning being "File ignored because no matching configuration was supplied."; (c) invariance — this repo's config never enables type-aware linting for any file (eslint.config.mjs, its own comment: "no parserOptions.project, no typed @typescript-eslint rules"), so nothing in this diff can move the verdict on a file it does not touch. No ablation: a one-field docs edit has no behaviour to mutate.

Serial probe before opening: git merge-tree --write-tree --name-only origin/main HEAD at origin/main = ffca0df9 printed a tree oid and no conflicted path (exit 0). git ls-remote --heads origin shows no other branch on this file.

Landing

⛔ Governed surface (docs/adr/**, Prime Directive #14) — draft, and it stays draft: no seat readies it, enqueues it, arms auto-merge, or submits a review on it. skip-changeset applied at open: this diff publishes nothing from any released package. content/docs/releases/** untouched. ADR-0126's **Original status**: history line untouched. ADR-0128 untouched.

Part of #15453, deliberately not a closing keyword: ADR-0128 is the card's third record and is still on the maintainer's A/B/C answer, so this PR must not close it.

验收备注

• noted, not filed — the ruling's execution note asks for os-zhuang + hotlong as requested reviewers. This seat opened the PR without requesting reviews, because on a governed surface a review request from a seat is the step the PM's ACCEPT stroke owns (the 速读 comment goes with it). Successor: the PM seat on this card's ACCEPT.
• noted, not filed — PR docs(adr): ADR-0131 — organization ownership is total: no NULL organization_id; declared metadata stays in code; rows only when an organization authored them (Refs #13564) #14976 reports merged_at 2026-09-04T05:37:45Z while its squash commit 0ed271574 carries 2026-09-04T04:50:16Z on both author and committer. Same calendar date, so the ruled date is unaffected either way; the ~47-minute gap is an observation about that one PR's landing, not a defect with a reproduction. Successor: none — nothing reads these two timestamps against each other.

维护者速读(草稿)

改了什么:ADR-0131 的状态行,从「Proposed —— 还等着你手工合并」改成「Accepted(2026-09-04)—— 落到 main 的那次合并就是受理动作」,并按裁决原话补一句:执行暂停由 #15193 和你的当面指令承载,不由这条状态行承载。整份记录只动这一个字段,别的一个字没动。

为什么改:这行说的是一件自我否证的事 —— 它把「你手工合并」称作待办,而读者能看到这行的唯一原因就是那次合并已经发生。它永远不会自己翻转,所以它不是状态,是噪声。你在 batch #40 回的「同意」就是对这三份记录的受理;本轮执行的是其中第二份(ADR-0130 已随 #15704 落地)。落地合并者实测是 os-zhuang(人),裁决要求的前提成立。

风险与代价(含回滚):主要风险是有人把「Accepted」读成「暂停解除了」—— 裁决为此专门加的那句话就写在同一行,而且 #15193 本轮实测仍是 open。回滚成本接近零:一个文件、一个字段、1 加 3 删,git revert 即可复原;这份 ADR 的状态行没有任何代码或门禁在读(scripts/adr-anchors/ 里没有指向 0131 的 anchor 文件,check-adr-symbol-anchors 本轮实测全仓「0 line anchors survive」)。第三份 ADR-0128 不在本 PR 里,它仍等你的 A/B/C。

席位意见:

你要做的:确认这个改法(尤其是「整个软换行字段一起换掉、连带删掉那句『在本记录合并前下面都不作数』」这个判断),然后手工合并 —— 受管面不进合并队列。ADR-0128 那张速读 5552091322 还在等你一个字母。

---

Generated by Claude Code

[os-musk]

维护者速读(终稿 · skills 席,2026-09-07T11:4xZ)

• 改了什么:ADR-0131 的状态字段从「Proposed —— 等你手工合并」改成「Accepted(2026-09-04)—— 落到 main 的那次合并(docs(adr): ADR-0131 — organization ownership is total: no NULL organization_id; declared metadata stays in code; rows only when an organization authored them (Refs #13564) #14976,0ed271574)就是受理动作」,并按你 batch Hide blog menu from top navigation #40 裁决的原话补一句:执行暂停由 [gate] the v18 development line is not open — ADR-0131 execution cards are blocked on this card #15193 与你的当面指令承载,不由这条状态行承载。整份记录只动这一个字段(1 加 3 删);ADR-0126 的历史行、ADR-0128、releases 文档全部未动。
• 为什么改:这行说的是一件自我否证的事 —— 它把「你手工合并」当待办,而读者能读到它的唯一原因就是合并已经发生;它永远不会自己翻转,所以是噪声不是状态,而 services 席一直把这行当暂停条件在读。ADR-0130 已按同一裁决落地(docs(adr): ADR-0130 status is Accepted, in ADR-0125's spelling, dated by its landing merge #15704),这是第二份;第三份 ADR-0128 因为是席位账号合并的,仍等你的 A/B/C([finding] Three merged ADRs still say "Proposed — awaiting the maintainer's hand-merge, which is itself the acceptance act" — a sentence their own presence on main disproves, and which seats read as a gate #15453 上的速读 5552091222)。
• 风险与代价(含回滚):风险是有人把「Accepted」读成「暂停解除」—— 裁决加的那句话就在同一行,[gate] the v18 development line is not open — ADR-0131 execution cards are blocked on this card #15193 本轮实测仍 open。回滚 = revert 一个 commit;零代码、零门禁读这行(check-adr-symbol-anchors 全仓「0 line anchors survive」)。
• 席位意见:席位亲核了三件事 —— PR docs(adr): ADR-0131 — organization ownership is total: no NULL organization_id; declared metadata stays in code; rows only when an organization authored them (Refs #13564) #14976 的 merged_by 是 os-zhuang(人),裁决前提成立;新行拼写与 ADR-0125 / ADR-0130 逐字同形;裁决那句逐字。「整个软换行字段一起换」与 ADR-0130 那次的做法一致,席位同意:留着「合并前下面都不作数」会让同一字段自相矛盾。CI 33 项在 11:36Z 无红(Lint & Repo Gates 与 Type Check · workspace 仍在跑,合并前会收敛)。建议直接合并。
• 你要做的:人工合并本 draft PR(受管面不进合并队列)。顺手给 ADR-0128 回一个字母(A/B/C)。

---

Generated by Claude Code