[os-elon-musk]

Requested by Elon · project thread

Fixes #32656

Stacked on #33276 (PR 1 of 2, base branch). Retarget to main once #33276 merges. The claim on #32656 names this branch.

Maintainer quick read

The refusals a permission admin or end user reads now come back in their language: request language, else workspace default, else English (ruling C on #32440). Codes, statuses and the diagnostic message are unchanged.

What changes

• plugin-security, packaged permission-set lock: PackagedPermissionSetLockedError and PackagedPermissionSetProvenanceUnknownError compose userMessage from an 11-language text table on the caller's ladder (each code matched by resolveBundleLocale, the platform's one locale ladder), instead of static English. The gate reads the ladder from its context; assertPermissionSetNotPackageDeclared and both constructors take it as an optional last argument (no locales ⇒ the English text, byte-identical to before).
• plugin-security, authz middleware: userFacingDenialMessage / fieldEditDenialMessage take [ctx.locale, ctx.workspaceLocale] and render in the first that has text (a deployment errors.<key> override, else the built-in operation-message catalog); none ⇒ en as before. Field labels are translated in the language the sentence resolves to.
• metadata-protocol (pass-through): MetadataAuthoringGateContext gains optional userTextLocales; runAuthoringGate forwards the save request's userTextLocales (from feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276) to the type's gate. Checked: E1's open PR fix(metadata-protocol)!: an App pair row declared for a package that still ships the name is never served at a bundle addition's address (ADR-0150 D4) #33251 touches none of these files.

Why

#32656 asks every userMessage producer to compose in the request locale with the declared fallback. #33276 did the save door's advisories and authoring-rule refusals; these are the remaining server producers that composed English only, or that read only the folded locale.

Verification

• packaged-permission-set-lock.test.ts: zh-CN, ja-JP (request wins over workspace), tlh→zh-TW workspace fallback, tlh/tlh = unstated = English; code/status/diagnostic identical in every language; the provenance-unknown refusal in de-DE.
• permission-denied-user-copy.test.ts: de-DE request with zh-CN workspace ⇒ zh-CN sentence; ja-JP request wins; de-DE/de-DE ⇒ en.
• mutation-listeners.test.ts: the gate receives the stated locales (empty entries dropped) and no key when none are stated.
• meta-save-user-text-locale.test.ts (runtime, both /meta doors): the permission gate receives ['zh-CN','de'], and [] with none.
• Local: @objectstack/plugin-security full suite 6610 passed; metadata-protocol / runtime targeted tests green; typecheck clean; test-source alias gate clean.

Clause-②: no
Metadata-protocol: none

Human Minimum

• 1. Schemas start as Zod; packages/spec holds no runtime logic.
• 2. Every name I added uses the casing for its kind.
• 3. No formatter ran; pnpm lint passes.
• 4. One issue, one branch off an up-to-date main, one pull request.
• 5. A change that publishes carries a changeset, and both declaration lines are filled in here and in the changeset.
• 6. packages/spec touched: its generated artifacts are regenerated and committed.
• 7. No hand edits to content/docs/references/, packages/*/CHANGELOG.md or content/docs/releases/.
• 8. Title, body and comments are in English.
• The Before Submitting list in CONTRIBUTING.md is done.

🤖 Generated with Claude Code

https://claude.ai/code/session_016WV4cz2MxDkQYCC4U8indM

[github-actions]

📓 Docs Drift Check

This PR changes 2 package(s): @objectstack/metadata-protocol, @objectstack/plugin-security, touching 25 documentable anchor(s).

1 hand-written doc(s) NAME something this change touched and may need an implementation-accuracy re-verification:

• content/docs/permissions/system-context.mdx (via registerAuthzMiddleware (symbol, a top-level function))

What this run could not see

• 1 name(s) were too generic to anchor anything (single lowercase words)
• the SDK route bridge reached 65 of 247 client-bound route-ledger rows — the other 182 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 182: 0 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 59 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 123 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 — 25 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 d683e6f3af5633a5c4a13ead523e06fbeb36db75 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 2987fb0cecac3eefc48c2d8ce813d8352fb2ba9f — the merge of head 24f2148f6bc4845bde87f441164c52c71eb8f35b into base d683e6f3af5633a5c4a13ead523e06fbeb36db75, 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 2987fb0cecac3eefc48c2d8ce813d8352fb2ba9f && git checkout 2987fb0cecac3eefc48c2d8ce813d8352fb2ba9f
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin d683e6f3af5633a5c4a13ead523e06fbeb36db75 24f2148f6bc4845bde87f441164c52c71eb8f35b && git checkout -B drift-repro d683e6f3af5633a5c4a13ead523e06fbeb36db75 && git merge --no-ff 24f2148f6bc4845bde87f441164c52c71eb8f35b

node scripts/docs-audit/affected-docs.mjs --json d683e6f3af5633a5c4a13ead523e06fbeb36db75

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

Advisory only, and a precision-first one (#9192): a page is listed because it names a
symbol, wire route or SDK method this diff touched — not because it mentions a changed
package. Each row says which anchor put it there, so a wrong row is reportable rather than
merely annoying. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs d683e6f3af5633a5c4a13ead523e06fbeb36db75 → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

Contract-review-claim: session_016WV4d1PGptBW3QRQNmy5e6 · head 0b4a03f · 2026-11-14T15:45:40Z

review#3 (area-lease pilot #33059, area A2).

[objectstack-fleet]

Contract-review-claim: session_016WV4d1PGptBW3QRQNmy5e6 · head fa8f2f5 · 2026-11-14T16:24:02Z

review#3.

[objectstack-fleet]

Review — PASS

Head: fa8f2f5d0a7bdd78250b80084ba3c9105a65322d · reviewer: review#3 (session_016WV4d1PGptBW3QRQNmy5e6) · area A2 · Fixes #32656 (PR 2 of 2, stacked on #33276 @caeb9046)

• Reviewed the diff over feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276's head (caeb9046...fa8f2f5d).
• Packaged permission-set lock. userTextLocales flows from the write door to MetadataAuthoringGateContext (filtered to non-empty strings) and on to assertPermissionSetNotPackageDeclared. Both refusals now pick their userMessage with resolveBundleLocale over the request locale, then the workspace default, then en. Texts cover all 11 PLATFORM_LOCALES. message (the diagnostic) and the status are unchanged.
• Permission middleware. Every userFacingDenialMessage call site now passes denialLocalesOf(opCtx.context) ([locale, workspaceLocale]). denialLocaleOf picks the first rung that has a deployment override or a built-in catalog entry, and returns undefined (so en) otherwise. A throwing i18n service is caught, so a 403 never becomes a 500. Field labels are translated in the language the sentence resolves to. Refusal decisions are untouched; only the language of userMessage moves.
• Clause-②: no holds: no accepted input, status or code changes; the user sentence's language now follows the ladder. Metadata-protocol: none holds.
• nit N1: the lock's 11 translations live in code, not in the operation-message catalog, so a deployment's errors.<key> override reaches the middleware's denials but not these two sentences. Acceptable for now; a catalog key would unify them later.
• Landing order: after feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276. Retarget this PR's base to main once feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276 merges, and re-read checks then.

CI was not complete when read (16:24Z); the ACCEPT re-reads every check at the landing head.

[objectstack-fleet]

Review carry — PASS stands

Head: 0a2173749d429d4d75ccce0f3e178c39abef47ed · reviewer: review#3 (session_016WV4d1PGptBW3QRQNmy5e6)

The delta from my PASS at fa8f2f5d0a7bdd78250b80084ba3c9105a65322d is the merge of #33276 at fedb4843 (that test-pin commit) and nothing else. No source line moves, so the PASS (and its nits) carries to this head. The ACCEPT cites this note as Regen-provenance and re-reads every check at the landing head.

[objectstack-fleet]

domain:server#5 · os-elon-musk · session_016WV4cz2MxDkQYCC4U8indM

ACCEPT — area A2 (permissions / audit / messaging / settings), area-lease pilot #33059.

• Fixes server: userMessage producers and the save door compose user-facing text in requestLocale (ruling C on #32440, layer 2) #32656 (2/2). PR 1/2, feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276, merged at 21:48Z; this PR's base was then retargeted from its branch to main.
• Reviewer: review#3 PASS (fix(plugin-security): permission refusals speak the caller's language, falling back to the workspace default (#32656, 2/2) #33290 (comment)), carried at 17:17Z (fix(plugin-security): permission refusals speak the caller's language, falling back to the workspace default (#32656, 2/2) #33290 (comment)).
• Delta since the carry, all merges, with no new source line:
  • 979fbb37 merges PR 1's branch.
  • 03adc6e8 merges origin/main after feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276 landed. Two conflicts: a comment line in protocol-publish-drafts-advisories.test.ts (took main's wording, as feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276 did), and an add/add in packages/runtime/src/domains/meta-save-user-text-locale.test.ts, where this PR's side extends main's (feat(metadata-protocol): a metadata write's advisories and refusals come back in the caller's language (#32656, 1/2) #33276) with the gated authoring-gate pin, so I took this PR's side. Its diff against main removes no line of main's except the three destructurings it widens.
  • 24f2148f is Regen-provenance: gen:schema and gen:docs on the merged tree; the os-regen hook reports both artifacts current. The PR's file list against main is now its own 12 files, and the generated pair is not among them.
• Local: meta-save-user-text-locale.test.ts 8/8 (after rebuilding metadata-protocol from this tree), protocol-publish-drafts-advisories.test.ts 11/11.
• Governed: check-governed-merges.mjs --pr 33290 reports NOT governed, 317 changed lines.
• Checks at 24f2148f: 51 judged, 35 success, 16 expected skips, 0 unexpected, 0 pending. mergeable_state is clean.
• K = 1 card.

Marking ready and enabling auto-merge.