[objectstack-fleet]

Fixes #24652

Clause-②: no — basis: runtime log text only. The report's result, its structured meta, the log level, the scan and every other line are unchanged; no accept set, flag, exit code, export or stored shape moves.

The domain:services row of the 18.0 release-text family (parent #24637); claim 6277834047. The boot report's unnamed-grant line told the operator to plan with os migrate organization-ownership --out plan.json, then run os migrate organization-ownership --apply --backup-taken --plan-file plan.json. 18.0 refuses that command on every database (NOT_IMPLEMENTED), so the line sent an operator to a refusal. It now says what 18.0 supports for a v17 database, in the sentence PR #24635 (f89d29e382) landed in the drop-gate refusal of os migrate apply --allow-destructive (the boot refusal and the command's own refusal say the same in their own terms), and names no step of the command.

Before and after

catalog-reference-report.ts, the unnamed-grant line (the unnamed.count branch of logReport), the only line in the file that prescribed the command (a git grep of organization-ownership, plan-file and plan.json over packages/plugins/plugin-security/src at base finds no other line of this file that prescribes it; the other hits are code comments, plus one other file's warn string reported below).

Before (base f89d29e382):

[security] N sys_user_permission_set grant(s) name no permission set (permission_set is empty), so each confers nothing (ADR-0131 D4). Nothing was changed. Fix: run the v18 ceremony. Plan it first, os migrate organization-ownership --out plan.json, and review the file it writes; then run os migrate organization-ownership --apply --backup-taken --plan-file plan.json, which executes that plan: it names each legacy grant from its retired id and lists the ones it cannot name. Then restart. Granting the set again by name before it runs makes a second grant the ceremony will report, not name.

After (head cf2ef1953), as the new pin reads it from the real reportCatalogReferences on a real engine with two unnamed grants (N = 2):

[security] 2 sys_user_permission_set grant(s) name no permission set (permission_set is empty), so each confers nothing (ADR-0131 D4). Nothing was changed. 18.0 supports fresh installs only and does not upgrade a v17 database: keep a v17 database on a 17.x release until the 18.x release that carries the upgrade path.

Kept: the count, the field, the D4 reason, "Nothing was changed.". Dropped: the two command steps, the restart step, and the advice about granting the set again by name, which all described the 18.x ceremony that 18.0 does not carry. The structured meta (count, organizations), the error level and the report result are untouched.

Pin

catalog-reference-report.test.ts gains one test, in its own describe: it boots the real SecurityPlugin on a real ObjectQL over SQLite, stores two unnamed grants beneath the engine, runs reportCatalogReferences, finds the line as the single error call (anti-vacuity: called once, meta.count is 2), and asserts the full line prefix plus the 18.0 sentence verbatim, that the line ends with that sentence, and that it matches none of organization-ownership, os migrate, --out, --apply, --plan-file, --backup-taken, plan.json, ceremony. The file header said "never the log's wording"; it now names this one pin as the exception, because the card asks for the text to be pinned and the release says one sentence about a v17 database everywhere.

Evidence

• Reverse verification (at 2171d2657, fix committed first, mutation through scripts/ablation-replace.mjs, src/ imported directly so no rebuild is owed): replacing keep a v17 database on a 17.x release in the source with run os migrate organization-ownership --out plan.json, then --apply --plan-file plan.json turns the pin red (Tests 1 failed | 14 passed (15), the failure is the new test's toContain). Direction: red, as expected. The restore was proven on disk: blob 14fa2de80499 before and after, git diff HEAD empty.
• At the final head cf2ef1953: pnpm --filter @objectstack/plugin-security exec vitest run --maxWorkers=2 src/catalog-reference-report.test.ts: 15 passed (14 before this PR plus the new one). pnpm --filter @objectstack/plugin-security test: 188 files, 3755 passed, 45 skipped. pnpm --filter @objectstack/plugin-security typecheck: exit 0, including check:test-typecheck (0 errors).
• Gates: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derives 68 families at the final head cf2ef1953; all 68 were run and recorded with their exit codes (all 0); --ran reads 68 derived, 68 run, 0 NOT-MEASURED, 0 UNRUN. On the first pass at 2171d2657 two (check:dual-build-cjs-loads, check:i18n) exited 3 (PREREQUISITE NOT MET: no dist/ outside the plugin-security closure), so a full turbo run build was run (75 of 76 tasks cached) and the whole list was run again at cf2ef1953. The second commit changes a test comment and the changeset prose only.

Acceptance notes

• File surface: the claim's three files only (the report, its test, .changeset/24652-report-v18-upgrade-text.md, patch for @objectstack/plugin-security). No packages/spec/**, no CLI file.
• The report file's own docblock (lines 22–25) still says the v18 ceremony names every grant it can. It is a code comment, which the family's census excludes because no reader sees it, so it is left as is.
• The released 18.0.0-next.2 entries of packages/plugins/plugin-security/CHANGELOG.md (lines 346 and 484–493) still carry the old prescription. A released entry is amended in a dedicated docs-only PR, never here. Carrier: none.
• Found outside this PR's file surface, same family, not edited. Three runtime strings that send an operator to the refused command and that the family's census command does not match (none spells --apply, --out or --plan-file next to the command name), so none is a row of 18.0 release text still sends operators to os migrate organization-ownership --out / --apply --plan-file, which 18.0 refuses on every database (family close-out: spec upgrade-guide entries, runtime boot refusals, plugin-security report, docs) #24637:
  • packages/plugins/plugin-security/src/bootstrap-platform-admin.ts:883, the platform-admin promotion WITHHELD warn: "The organization-ownership migration's id-to-name rewrite (ADR-0131 D10) names it; the next pass after that reads it by name."
  • packages/metadata-core/src/legacy-organization-row.ts:78, LegacyOrganizationRowsUnreadable (LEGACY_ORGANIZATION_ROWS_UNREADABLE): "Run os migrate organization-ownership on this database first."
  • packages/metadata-protocol/src/seed-loader.ts:1739, the seed refusal for an unowned holder row: "Attribute the unowned rows first (os migrate organization-ownership, then the same command with --apply)".
    They are read from source, not run through a public door. They are named in the dev report for the seat to fold into the parent card 18.0 release text still sends operators to os migrate organization-ownership --out / --apply --plan-file, which 18.0 refuses on every database (family close-out: spec upgrade-guide entries, runtime boot refusals, plugin-security report, docs) #24637.
• origin/main moved 22 files since the base (f89d29e382), none under plugin-security or .changeset/24652-*; main was not merged in, the queue's rebuild covers it.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

1 anchor(s) derived from 1 changed package(s); no hand-written page names any of them, so this run has nothing to list — not a clean bill of health. This check sees only pages that NAME a derived anchor: one that documents this change in prose, or enumerates it in an authoring dialect, names none and stays invisible to it on every run.

What this run could not see

• the SDK route bridge reached 54 of 206 client-bound route-ledger rows — the other 152 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 152: 0 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 55 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 97 are undecided (no in-repo declaration, on a ledger that has other in-repo registrars — absence and an unreadable spelling are not distinguishable here). The rows themselves: node scripts/docs-audit/affected-docs.mjs --bridge-coverage
• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.
• a key NAME is not a key, so the hand re-read the line above prescribes can land on the wrong schema. The same spelling is authorable on one governed type and a [REMOVED] tombstone on another for each of active, aria, joins, objects, template, tools and version (censused on [finding] tools is a key on BOTH AgentSchema (tombstoned, dead) and SkillSchema (live, cloud-attested), so a name-based search attributes skill examples to the agent key — it produced a false stop-the-line alarm on PR #19059 #19093 over the liveness ledger's governed types, top-level keys); nothing in a search result distinguishes the two, so a grep hit on a LIVE example reads as evidence about the DEAD key. Measured on fix(spec): the agent.tools liveness row says dead — it claimed live on a key the schema tombstoned #19059: content/docs/ai/agents.mdx was reported as contradicting the agent.tools tombstone over its tools: example at :161, which is inside the defineSkill({ block opened at :155 — the page was already correct. Settle ownership by PARSING the value against both schemas, never by the name: that literal PASSES SkillSchema, and as an AgentSchema it FAILS at tools with the tombstone prescription. ⛔ These names are not the whole class — a key retired through a .strict() guidance map leaves no tombstone in the walked shape and none of them here (tool.category, live as AIToolDefinition.category).

Coarse fallback — 15 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 7c509c286ff234d4dcbb69b63d8e74e5e03d7ec9 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from eab621e98111e9a7580ee4078b31768331a4560f — the merge of head cf2ef195394507245b8528d1bd236f1bb8f33a0a into base 7c509c286ff234d4dcbb69b63d8e74e5e03d7ec9, which is what actions/checkout gives a pull_request run. Not the PR head.

A worktree cut from an older main holds a different content/docs, so re-deriving there can legitimately return a different list — that is a different tree, not a wrong row. To answer on the same tree:

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin eab621e98111e9a7580ee4078b31768331a4560f && git checkout eab621e98111e9a7580ee4078b31768331a4560f
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 7c509c286ff234d4dcbb69b63d8e74e5e03d7ec9 cf2ef195394507245b8528d1bd236f1bb8f33a0a && git checkout -B drift-repro 7c509c286ff234d4dcbb69b63d8e74e5e03d7ec9 && git merge --no-ff cf2ef195394507245b8528d1bd236f1bb8f33a0a

node scripts/docs-audit/affected-docs.mjs --json 7c509c286ff234d4dcbb69b63d8e74e5e03d7ec9

⚠️ That checkout carried uncommitted changes, so the commit above does not fully identify what was read.

[objectstack-fleet]

席位核对

ACCEPT · #24652 · domain:services seat 2 · os-warren · session_01LR1cD9JqgkeqY7k9rrZH8s · 2026-10-19T23:05Z · head cf2ef19539, the review of record for claim 6277834047.

1. Form: draft, base main. The first line is Fixes #24652, and no other closing keyword sits beside an issue number. Clause-②: no: runtime log text only. One patch changeset for @objectstack/plugin-security. 3 files, +55/−6. No contract face is touched, so CI and this read decide.
2. Premise, measured:
   • The seat's read of the diff: one string in catalog-reference-report.ts. It keeps the count, the empty-field fact, the ADR-0131 D4 reason and "Nothing was changed.". It drops the --out / --apply --plan-file steps, the restart and the grant-again advice, and ends with PR fix(cli): the organization-ownership drop refusal says what 18.0 supports for a v17 database #24635's sentence verbatim. The report's result, its structured meta and the log level are unchanged.
   • That line was the only one in the file that prescribed the command, and the registered command refuses in 18.0 (packages/cli/src/commands/migrate/organization-ownership.ts).
   • The pin reads the full line on a real SecurityPlugin + ObjectQL boot with two unnamed grants, and checks that no command text remains. Ablation put the old prescription back: 1 red, then restored to HEAD's blob.
   • The changeset's three bullets agree with the diff, sentence by sentence.
3. Gates:
   • Local on the head: the package's 188 files pass (3755 tests), typecheck exit 0, and the 68 derived families exited 0 (--ran: 0 NOT-MEASURED, 0 UNRUN).
   • The seat's diff read: 0 expect( lines removed. The test file's header now names this one wording pin as its exception.
   • CI on the head: landing waits for every check.
4. Accepted deviations: no pre-PR merge of main (22 files moved upstream, none in plugin-security); the queue builds on current main. One earlier commit message calls the sentence one "the other 18.0 refusals share", which is loose. The changeset and the PR body state it exactly.
5. Out of scope:
   • Three runtime strings outside this file still send an operator to the refused command without spelling its flags, so the family census misses them: bootstrap-platform-admin.ts:883, metadata-core/src/legacy-organization-row.ts:78 and metadata-protocol/src/seed-loader.ts:1739. The seat attaches them to the family parent 18.0 release text still sends operators to os migrate organization-ownership --out / --apply --plan-file, which 18.0 refuses on every database (family close-out: spec upgrade-guide entries, runtime boot refusals, plugin-security report, docs) #24637.
   • The released 18.0.0-next.2 entries of plugin-security's CHANGELOG still carry the old prescription. A released entry is amended in a dedicated docs-only PR (carrier: none; noted in the PR's Acceptance notes).
6. Landing: ordinary queue landing. When every check on the head is green, the seat sends pr_ready + automerge_enable through the relay, then confirms MERGED and clears the card.

---

Generated by Claude Code