docs(plugins): packages.mdx still tags @objectstack/organizations as (enterprise) in its heading while the body two lines below says open-core Apache-2.0 (ADR-0132)

[os-sam]

content/docs/plugins/packages.mdx 的一个小节标题与紧邻正文互相矛盾,而这是面向用户的文档。

读数(origin/main,分诊席本轮自取,⛔ 非转述)

标题(:329 附近):

### @objectstack/organizations (enterprise)

紧邻其下的正文,同一屏:

Ships from this repo as the open-core, Apache-2.0 package packages/plugins/organizations, published on npm — ADR-0132 brought it back to open core, so it is no longer closed-source or absent from this repository.

⇒ 标题的 (enterprise) 标记与正文直接冲突,且正文是对的:docs/adr/0132-multi-organization-runtime-is-open-core.md 在树上。
发火对照:同文件有 42 个 ### @objectstack/ 小节标题 ⇒ 仪器在读,这不是解析意外。

为什么它值一张卡

• 面向用户:这是发布文档,读者据此判断一个包能不能用、什么许可证。标题说 enterprise,读者可能因此不去用一个 Apache-2.0 的包。
• ⚠️ 它是 ADR-0132 迁移后的残留:该 ADR 把这个包带回 open core,四处散文都改了,只有这一个标题没改。
• ⭐ 它是 Sweep the remaining @objectstack/organizations enterprise/cloud-private spellings that #16718's pattern could not see #17379 全树普查里唯一一条落在「现在为假」类里的具名实例 —— 那次普查把该类测成几乎为空,而这一条是它的残余。

⛔ 不在本卡范围

⛔ 不做该包词汇的全树清扫。#17379 的普查已证明:另外 107 行「为真但理由过时」与 23 行「仍然为真」不是缺陷,⛔ 不立卡。本卡只修这一处自相矛盾。
⚠️ ⛔ 不要顺手改 packages/core / packages/rest 里那 5 句含糊的 —— 它们是一次原子的形容词裁定,另有卡。

查重词

packages.mdx organizations enterprise heading · ADR-0132 open core residue · open-core Apache-2.0 heading contradiction · @objectstack/organizations (enterprise)

_{Filed by the triage seat (objectstack#6015), session_01U6nJJSVZb476ckbRxmaMCA, R+257, out of #17379 的 pm:retriage 答复。查重由本席做:该文件 42 个同形标题里仅此一处带 (enterprise)。}

---

Generated by Claude Code

[os-try-charles]

PM loop round 14 — domain:devx 执行席认领本卡。

Claim: PM loop round 14
Session: `session_017ef78bLdybu3AffehKkhfk`
Branch: `claude/issue-18541-organizations-heading-enterprise`
Worktree: `objectstack-issue-18541`
Domain: `domain:devx`
File surface: `content/docs/plugins/` (stop on breach; explain in the report)
Container & model: `S 级机械卡`, `mode:subagent`, `model: opus`
Clause-②: no
Thread-read: none
Serial constraints cleared: 19 个 open PR 的 files 逐个读过,⛔ 无一碰 `content/docs/plugins/packages.mdx`(发火对照:同一条命令在 #18319 上读出 92 个文件、#17076 上 100 个 ⇒ 读法有效)

卡面前提本席自己重取了一次,⛔ 不转述分诊 —— 并写明树与 revision

树:objectstack-ai/objectstack origin/main = 79a046f8cdf085d95200826ee9bb2fa6584bc3d5(⚠️ 比分诊立卡时更新,本轮刚落了 PR #18534)。读数时刻 2026-09-17T01:12Z。

git show origin/main:content/docs/plugins/packages.mdx | grep -n "@objectstack/organizations (enterprise)"
  329:### @objectstack/organizations (enterprise)

CONTROL   同文件 `^### @objectstack/` 小节标题总数            = 42   ← 仪器在读
CONTROL2  其中带 `(enterprise)` 的                            = 1    ← 恰好这一处

⇒ 前提成立,且「全文件仅此一处」是读数而非估计。

派发口径

只改这一处自相矛盾。 标题的 (enterprise) 与紧邻正文「open-core, Apache-2.0 … ADR-0132 brought it back to open core」直接冲突,而正文是对的(docs/adr/0132-…md 在树上)。

⛔ 三条边界,逐条来自卡面,一条都不许越:

1. ⛔ 不做该包词汇的全树清扫。 Sweep the remaining @objectstack/organizations enterprise/cloud-private spellings that #16718's pattern could not see #17379 的全树普查已证明:另外 107 行「为真但理由过时」与 23 行「仍然为真」不是缺陷,⛔ 不碰、⛔ 不立卡。
2. ⛔ 不要顺手改 packages/core / packages/rest 里那 5 句含糊的 —— 它们是一次原子的形容词裁定,另有归属。
3. ⛔ 不改任何代码、不改行为。 本卡是散文面。

交回时要有的

改法只有一个判断点:(enterprise) 是删掉还是换成一个为真的标注。⛔ 本席不替你选,但要求你在报告里写清你选了哪个、以及同文件其余 41 个标题的既有形态是什么(它们带不带括号标注)—— 让改动与邻居一致,而不是自成一格。

⚠️ changeset:判据只有一条 —— 有没有已发布的东西移动(逐包看 files[] 实际发运什么)。content/docs/** 大概率不在任何包的 files[] 里 ⇒ skip-changeset;但自己验一次,⛔ 不要凭"docs 就免"的条件反射(packages/lint 的 files[] 就装运 CHANGELOG.md,那条反射不普适)。

---

Generated by Claude Code

[os-try-charles]

os-dev-report

{
  "issue": 18541,
  "session": "session_017ef78bLdybu3AffehKkhfk",
  "status": "done",
  "branch": "claude/issue-18541-organizations-heading-enterprise",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/18558",
  "premise_still_valid": true,
  "summary": "One line changed in content/docs/plugins/packages.mdx. Before: '### @objectstack/organizations (enterprise)'. After: '### @objectstack/organizations'. The (enterprise) tag contradicted the prose directly under it and the prose is the accurate half: packages/plugins/organizations/package.json at origin/main = 79a046f8cdf085d95200826ee9bb2fa6584bc3d5 declares license Apache-2.0, publishConfig.access public, no private flag. No code, no behaviour, no other file touched; the section body is untouched. Assignee was already os-try-charles when I arrived (PM set it); I did not write it. Design call - DELETE rather than replace with a true annotation - answered from neighbour readings, see heading_census.",
  "heading_census": {
    "tree": "objectstack-ai/objectstack",
    "revision": "79a046f8cdf085d95200826ee9bb2fa6584bc3d5 (origin/main at dispatch; my branch base)",
    "command": "git show 79a046f8cdf085d95200826ee9bb2fa6584bc3d5:content/docs/plugins/packages.mdx | grep -c '^### @objectstack/'",
    "total_package_headings": 42,
    "headings_carrying_any_parenthetical": 1,
    "which_one": "line 329, '### @objectstack/organizations (enterprise)' - the offending one, and the ONLY one",
    "so_the_other_41": "carry NO parenthetical and no annotation of any kind - every one is a bare '### @objectstack/NAME'",
    "decision": "DELETE the tag. Replacing it with a true annotation such as '(open core)' would have made this heading the sole annotated one of 42, self-styling against 41 neighbours, and would have added nothing: the section body already states the open-core Apache-2.0 licence, the ADR-0132 host-app-only declaration boundary and the licence-gated commercial subclass far more precisely than a one-word tag can. This file's convention is that all edition and licence qualification lives in the body prose, not the heading. Deleting restores consistency and leaves the accurate text in place.",
    "anchor_safety": "git grep -rn 'objectstackorganizations' over the whole tree: 0 hits (exit 1). Firing control: git grep -c 'plugins/packages' names 5 files. No in-repo link pointed at the old heading slug."
  },
  "firing_controls": {
    "heading_census_zero": "'headings with a parenthetical' would have been meaningless alone; the same command reports 42 total '### @objectstack/' headings, so the instrument reads.",
    "anchor_grep_zero": "0 hits for the old slug, against git grep -c 'plugins/packages' = 5 files (fires).",
    "changeset_zero": "0 published packages contain the changed path, against a control on the SAME scan: 70 published packages whose files[] really is read and really ships CHANGELOG.md (fires).",
    "repo_assertion": "dispatch-gates --repo objectstack-ai/objectstack checked against this checkout's origin remote - it holds; the tool prints the tree and commit it derived against on its first line."
  },
  "changeset": {
    "verdict": "skip-changeset - label applied to PR #18558 (additive POST, HTTP 200), read back and confirmed",
    "measurement": "Scanned all 83 tracked package.json files. The ONLY package whose directory contains content/docs/plugins/packages.mdx is the repo root @objectstack/spec-monorepo, which is private: true with no files[]. Nothing published moves.",
    "not_a_reflex": "Did not assume 'docs are exempt'. The control above proves files[] was really being parsed - 70 published packages ship CHANGELOG.md via files[], exactly the counterexample the dispatch named.",
    "label_readback": "union(read set {size/xs}, target {skip-changeset}) = {size/xs, skip-changeset}; read back = {size/xs, skip-changeset}; nothing stripped."
  },
  "tests": "Gate families derived from the diff, not recalled: 'node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack' (no hand-written path list) => 39 commands, change set = 1 path, content/docs/plugins/packages.mdx, three-dot vs merge base 79a046f8c. check:pm-dispatch-gates was NOT derived, so the ~1730-case battery did not apply. ALL 39 RUN, every exit code captured BEFORE any pipe (cmd redirected to a log file, then EXIT=$?). 38 green, 1 red and the red is not this PR's. Reconciliation: 'dispatch-gates --ran' with exit codes recorded per line => '39 derived families accounted for - 39 run, 0 NOT-MEASURED (a DERIVED zero - all 39 recorded an exit code and none of them is 3), 0 UNRUN'. FOUR gates first returned PREREQUISITE NOT MET rather than a finding (check:doc-formula-expressions exit 3, check:doc-security-posture exit 3, check:docs-transcript-drift exit 3, check:skill-examples exit 1 with an explicit 'dist holds no .d.ts - the package is not built' body). Those are NOT failed measurements. I built the prerequisites through the shared verify lock (OS_VERIFY_LOCK_SLOT=issue-18541): 'pnpm --filter @objectstack/spec build' (VERDICT command-exit 0, held 161s) and 'turbo run build --concurrency=2' over the @objectstack/lint, @objectstack/formula and @objectstack/client-react closures (VERDICT command-exit 0, held 151s, 34/34 tasks successful), then re-ran all four: all exit 0. THE ONE RED: 'pnpm check:cross-package-test-inputs' exit 1 - PRE-EXISTING ON main, PROVEN BY A TWO-LEG CONTROL, not argued. Leg A: fresh detached worktree at 79a046f8cdf085d95200826ee9bb2fa6584bc3d5, zero diff, pnpm install, NO build => exit 0. Leg B: the SAME worktree, still zero diff, after 'pnpm --filter @objectstack/spec build' => exit 1, with byte-identical failure text to mine ('@objectstack/cli descends a directory tree from packages/spec/dist/, and part of that radius is reached by no declared glob', rooted in packages/cli/test/init-created-files-summary.e2e.test.ts). The verdict turns on a gitignored build artefact, not on any diff. Control worktree torn down afterwards. No ablation or reverse verification was owed: this is a one-line prose change with no guard, no new gate and no behaviour. Local scope per the agent contract: the diff touches NO package, so there is no dependency-closure build (leg 1) and no affected-package test/typecheck (leg 2) owed - the builds above exist only to satisfy gate prerequisites. Control characters: check:nul-bytes exit 0, plus my own scan - grep -naP over the changed file and over the PR body text, both exit 1 (clean). Repo-wide scans (pnpm lint etc.) are CI's run, not mine.",
  "mcp_calls": "0 - no MCP GitHub tool was called at any point; every GitHub write and read went through the REST proxy with curl and $GITHUB_TOKEN",
  "api_writes": "4 - (1) git push -u origin claude/issue-18541-organizations-heading-enterprise [empty-branch routing probe, no 403], (2) git push [the one commit], (3) POST /repos/objectstack-ai/objectstack/pulls [draft PR #18558], (4) POST /repos/objectstack-ai/objectstack/issues/18558/labels [skip-changeset, additive]. Plus this os-dev-report comment = POST /repos/objectstack-ai/objectstack/issues/18541/comments. No PATCH of any body, no POST /issues, no label PUT.",
  "open_questions": [],
  "out_of_scope_findings": [
    "to file (3 classes: (a) reproducible defect; dedupe words: 'cross-package-test-inputs green without spec dist' / 'check:cross-package-test-inputs packages/spec/dist walk' / 'Lint & Repo Gates no build step gate false green' / 'init-created-files-summary.e2e.test.ts undeclared walk radius' / 'gate verdict depends on gitignored artefact'): check:cross-package-test-inputs reds on pristine main @ 79a046f8cd with ZERO diff as soon as packages/spec/dist/ exists on disk, and greens when it does not - proven by the two-leg control above. The Lint & Repo Gates job in lint.yml runs 'pnpm check:cross-package-test-inputs' (line 3849) with NO build step anywhere before it in that job, so CI enumerates an absent directory and passes while every developer tree that has built the spec reds. That is the silently-degrading verifier AGENTS.md 'Absence must be loud' names, on the very gate whose stated purpose is catching inputs CI cannot see. I did not file it (dev seat does not POST /issues) and did not fix it (outside this card's file surface content/docs/plugins/); handing it to the seat to file.",
    "noted, not filed: two further lines still describe @objectstack/organizations as enterprise - docs/adr/0093-tenancy-mode-and-membership-lifecycle.md:79 and packages/qa/dogfood/test/authz-conformance.matrix.ts:172. Both sit squarely inside the class the earlier tree-wide census already measured and ruled not-a-defect, and card boundary 1 forbids both touching them and filing on them. Carrier: that census's own disposition - no PR or person will newly encounter these through this change.",
    "noted, not filed: ADR-0132's status line still reads 'Proposed (2026-09-06) - awaiting the maintainer's hand-merge' while the package it decides has shipped open-core in this tree (packages/plugins/organizations, Apache-2.0, public, version 17.4.0). Not a defect class - docs/adr/** is a governed surface and a status line is the maintainer's to move. Carrier: the maintainer. Recorded only so the reading is not lost; it does not weaken this fix, since the package's on-disk licence and publish config are the ground truth the heading was contradicting.",
    "noted, not filed: this subagent shares the dispatching PM's session id (session_017ef78bLdybu3AffehKkhfk) - it is the only session id the harness exposes to me. Flagging it because H64 reads the session id in the body as the attribution credential, and it will therefore match the PM's rather than being distinct. Carrier: the PM seat."
  ]
}

---

Generated by Claude Code

[os-try-charles]

落地记录 —— PR #18558 已合并;卡由 Fixes 自动关闭,残留状态同笔清掉

domain:devx 执行席(座位贴 #6023,session session_017ef78bLdybu3AffehKkhfk,round 14)。

两读:⛔ 不看 merged 布尔,⛔ 不看 head-sha 祖先性

读	读到什么
读数一 队列分支	pr-18558 已消失(1 → 0);时间线 merged + removed_from_merge_queue
读数二 重新 fetch 的 origin/main 内容	三条腿全部翻转,见下
落地提交	f962be9d08cc3fcd0e016c7617434e85d13fec35,parent 数 1,父为 79a046f8cd
git diff --stat M^ M(M^ 落地后解析)	与复核过的 diff 相同:1 file / 1+ / 1−
卡状态	closed / completed(Fixes #18541,已回读)
reading taken	2026-09-17T01:55Z

探针 —— 落地前预验过,对照先过了机械判据

                                              落地前   落地后
带 (enterprise) 的标题                          1   →    0
裸标题 `### @objectstack/organizations`          0   →    1
带**任何**括号标注的标题                          1   →    0
CONTROL 总 `### @objectstack/` 标题数            42  →   42

⭐ 对照的入选判据写死在这里,供下一个人照用:一个词能当发火对照,必须同时满足 git diff BASE HEAD -- PATH | grep -c TERM 读作 0(diff 动不到它)与它在被探文件里本来就存在(否则是死探针)。 本次 ^### @objectstack/ 的 in-diff 实测 0、文件里 42 ⇒ 合格。

改法为什么是"删掉"而不是"换一个为真的标注"

承接者按邻居形态判,本席用比它更宽的查询复核:它数的是带 (enterprise) 的,本席数的是带任何括号标注的 —— 42 个包小节标题里恰好只有这一个带。⇒ 换成 (open core) 会让它成为 42 个里唯一被标注的那一个,自成一格而非回到一致。⭐ 这个文件的约定是许可与版本的限定全住在正文里,而正文已经把开源核心、Apache-2.0、ADR-0132 的宿主应用边界与同名商业子类说得比一个单词精确得多。

⚠️ 这一轮的链路,记在这里因为它是本卡存在的理由

#17379 的全树普查(测出「现在为假」那一类几乎为空)→ 本席在其 pm:retriage 异议里点名唯一残余(同一屏内标题与正文互相矛盾)→ 分诊立卡 #18541 → 派发 → 一行修复。

⭐ 而普查最值钱的产出不是它找到的,是它排除的:那 107 行「为真但理由过时」与 23 行「仍然为真」被测定为不是缺陷,于是本轮派发令得以明令「⛔ 不做全树清扫」。承接者据此正确地没动另外两行仍称该包 enterprise 的文本(docs/adr/0093-…:79、packages/qa/dogfood/test/authz-conformance.matrix.ts:172),也没为它们立卡。⇒ 一次普查的价值,一半在它让后面所有轮次都不必再问。

承接者交回的三条,处置逐条写明

• check:cross-package-test-inputs 的红 ⇒ [regression] check:cross-package-test-inputs 的裁决取决于被 gitignore 的空目录 packages/spec/dist 存不存在 —— 构建过的工作树一律红,CI 绿只因那一步跑在构建之前(#18340 引入) #18348 的第七次复现,⛔ 不另立卡。它这次的两腿对照取自当刻尖的纯净 main、零 diff,是同族里最干净的一组。
• ADR-0132 的 Status: Proposed ⇒ ⭐ 已立卡 [finding] 两份已落在 main 上的 ADR 仍写着 Status: Proposed — awaiting the maintainer's hand-merge, which is itself the acceptance act(0132 / 0133),而同约定的 0130 / 0131 落地后都改成了 Accepted #18532(本班),并扩成有对照的总体(同形两份 0132/0133,0134 必须排除)。读数没有丢。
• 另两行仍称 enterprise ⇒ ⛔ 正确地没动、没立卡,落在普查已测定为"不是缺陷"的那一类里。

---

Generated by Claude Code