[objectstack-fleet]

Fixes #24937

Clause-②: no

What was wrong

A turso, mongodb or sqlite-wasm datasource whose optional driver package is not installed is refused at boot. The factory's own text already says so and names the install (npm install @objectstack/driver-turso, with the pnpm add and yarn add spellings beside it). The refusal then ended on UNBUILT_WORKSPACE_REMEDY anyway: "… That is a build precondition … Run pnpm install && pnpm build, then start again." That is two fixes that contradict each other, and the second does nothing for a project, which adds these packages itself.

connectFailureRemedy() chose the unbuilt remedy for every module-not-found except the three database clients PR #24933 split off (pg, mysql2, better-sqlite3). A bare @objectstack/driver-* specifier still classified as unbuilt, and four factory classifier pins asserted that.

What changed

Seat ruling B on the card. The code change is in packages/services/service-datasource/src/connect-failure-remedy.ts, and it is a message branch only.

• New closed list OPTIONAL_DRIVER_PACKAGES, built from the factory's own constants (TURSO_DRIVER_PACKAGE, MONGODB_DRIVER_PACKAGE, SQLITE_WASM_DRIVER_PACKAGE). A pin holds it equal to the optional peers package.json declares (peerDependenciesMeta.*.optional). Measured set: exactly those three.
• missingOptionalDriverPackage(cause) reads the same first Cannot find module|package 'X' specifier PR fix(service-datasource): a missing database client (pg, mysql2, better-sqlite3) is refused with its install, not the unbuilt-workspace build advice #24933 introduced. It answers X only when X is one of those three, as a bare name.
• connectFailureRemedy() then ends on missingOptionalDriverPackageRemedy(X). That is ONE sentence deferring to the install the factory text already names: "The fix is the install of X named above, then start again: it is an OPTIONAL driver package the project adds itself, so a build cannot produce it and no boot-time override can stand in for it." It names no command, no build, no environment variable and no configuration key.
• isUnbuiltWorkspaceFailure() answers false for that case.
• These keep their remedies:
  • A PATH into one of those packages (…/driver-turso/dist/index.mjs) keeps UNBUILT_WORKSPACE_REMEDY, and so does a subpath of one.
  • @objectstack/driver-sql and any other @objectstack/… name keep it too.
  • The three clients keep PR fix(service-datasource): a missing database client (pg, mysql2, better-sqlite3) is refused with its install, not the unbuilt-workspace build advice #24933's remedy.
  • Every other cause keeps GENERIC_CONNECT_FAILURE_REMEDY, byte for byte.
• The module note gains a section for this case.

Unchanged, per the hard stops:

• The factory's own messages: default-datasource-driver-factory.ts is not in the diff.
• When boot is refused, the error type, the retained state and the OS_ALLOW_DRIVER_CONNECT_FAILURE path.
• sqlite-driver-fallback.ts, packages/types/**, packages/drivers/** and packages/spec/** are not touched.
• No new env var, flag or gate.

Premise re-check (the ruling's P), measured before any edit

P: in this monorepo, once pnpm install has run, a missing or unbuilt optional driver surfaces as a PATH, not as a bare Cannot find package name.

Setup: a fresh worktree at 8cfe1f94cb, pnpm install --frozen-lockfile, no build. All three packages are linked into service-datasource/node_modules/@objectstack/.

• Resolver leg (no build at all): Node 22, from packages/services/service-datasource. For each package, ESM and CJS both answer a path:
  • ESM: ERR_MODULE_NOT_FOUND | Cannot find module '…/service-datasource/node_modules/@objectstack/driver-turso/dist/index.mjs' imported from …
  • CJS: MODULE_NOT_FOUND | Cannot find module '…/driver-turso/dist/index.js'
  • driver-mongodb and driver-sqlite-wasm answer the same shape.
• Boot leg: the service's closure built EXCEPT the three drivers, then a datasource bound to my_app_note booted per driver through the real factory and the real DatasourceConnectionService (built dist/). Every arm's first unresolved specifier is the …/driver-X/dist/index.mjs PATH.

P holds, so B ships. A contributor with an unbuilt checkout is still told to build. That is unchanged by this PR, and the pins for it are now path-shaped.

Measured on the card's path, before and after

The published @objectstack/service-datasource@18.0.0 was installed alone in an empty directory, with no driver packages (npm install --ignore-scripts). A declared datasource bound to one object was booted per optional driver through the published factory and DatasourceConnectionService, under both dist/index.js (ESM) and dist/index.cjs (CJS).

• Before (18.0.0):
  • the first unresolved specifier is the bare @objectstack/driver-turso (-mongodb, -sqlite-wasm);
  • the refusal names npm install once AND ends on "… Run pnpm install && pnpm build, then start again."
• After: the same directory with this branch's packed build installed over the published one (pnpm pack of service-datasource, marker grep: positive control 1 hit, the 18.0.0 copy 0 hits):
  • all three arms, ESM and CJS, end on "The fix is the install of @objectstack/driver-X named above, then start again: …";
  • npm install appears exactly once in the whole refusal;
  • pnpm install and pnpm build are absent.
• NOT MEASURED: the os dev scaffold path itself (create-objectstack@18.0.0, a turso datasource). Reason: the CLI reaches this refusal through the same factory and DatasourceConnectionService measured above. The card measured the scaffold's before-state, and a scaffold after-state needs this unpublished build overridden into a full CLI install.

Tests (head 87b717e5a0)

• pnpm --filter @objectstack/service-datasource typecheck: tsc --noEmit, exit 0. --listFiles includes both edited test files.
• pnpm --filter @objectstack/service-datasource test: Test Files 49 passed (49), Tests 899 passed (899).
• New pins (datasource-connection-service.test.ts):
  • Each optional package × {ESM Cannot find package 'X' imported from … with ERR_MODULE_NOT_FOUND, CJS Cannot find module 'X' plus Require stack with MODULE_NOT_FOUND} goes through the factory's real message builders (the turso one as MissingDriverPackageError). Each ends on the deferring sentence. The whole refusal holds exactly one npm install, no pnpm build, no pnpm install, no unbuilt remedy and no generic advice.
  • A path into each package keeps the unbuilt remedy, and so does a subpath.
  • The remedy is one sentence with no command, OS_* variable or --database.
  • The escape hatch still boots degraded.
  • The list equals the manifest's optional peers.
  • Classifier units: bare name only, module-not-found only, @objectstack/driver-sql is not on the list, a code-only error names nothing.
  • The existing fix(service-datasource): a missing database client (pg, mysql2, better-sqlite3) is refused with its install, not the unbuilt-workspace build advice #24933 client pins and the ECONNREFUSED / 28P01 generic pins are unchanged and green.
• Revised factory classifier pins (default-datasource-driver-factory.test.ts). The card says three; there are FOUR sites asserting isUnbuiltWorkspaceFailure(...) true on a bare-package cause: the turso, sqlite-wasm and mongodb "keeps the import error verbatim" cases, and the three-arm parity case.
  • Each now asserts the unbuilt reading on a PATH-shaped cause (unbuiltCause(pkg), the measured shape).
  • Each asserts the bare name is that package's missing install.
  • The parity block gains one case for the bare reading.
• No real-path pin was added, because none is cheap here. vitest wraps every mock-factory throw in its own "[vitest] There was an error when mocking a module" error (@vitest/mocker createHelpfulError, measured in the installed 4.1.11). A staged absence therefore cannot carry Node's text to the classifier. The factory has no import seam, and its file is off this PR's surface. The before/after run above is the real-path evidence instead.

Ablation (one leg per changed behaviour; fix committed first; node scripts/ablation-replace.mjs, WRAP mode)

Each leg ran the three suites datasource-connection-service, default-datasource-driver-factory and missing-driver-client-boot-refusal. The subjects resolve through relative src imports, so no dist/ leg applies.

1. Delete the connectFailureRemedy branch (if (driverPackage !== undefined) return missingOptionalDriverPackageRemedy(driverPackage);).
   • On disk: anchor 1 → 0, blob e6403dd54e18 → bf36838df528.
   • Predicted red. Result: Tests 6 failed | 150 passed, exactly the six ESM/CJS deferring-remedy pins.
   • Restored by git checkout HEAD --: blob e6403dd54e18 == HEAD, git diff HEAD empty, git status clean.
2. Drop the isUnbuiltWorkspaceFailure exclusion (&& missingOptionalDriverPackage(cause) === undefined).
   • The first attempt was a no-op. The tool refused it on its own replacement-count check (newline replacement), restored, and never ran the suites.
   • Re-run with a distinctive replacement. On disk: anchor 1 → 0, replacement 0 → 1, blob e6403dd54e18 → ebf06e9d96f4.
   • Predicted: the classifier pins go red, and the service-level remedy pins stay green because the branch runs first. Result: Tests 5 failed | 151 passed, exactly the five classifier pins. The four factory sites plus the unit failed; the remedy pins stayed green.
   • Restored: blob == HEAD, git diff HEAD empty.

Gates (local subset, head 87b717e5a0)

• Derivation: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack, 66 commands, 4 changed paths vs merge base 8cfe1f94c.
• Each command was run once, exit captured before any pipe.
• Result: 65 exit 0, and 1 NOT MEASURED, pnpm check:dual-build-cjs-loads. It exited 3 with PREREQUISITE NOT MET: it reads every package's dist/, and this worktree builds only the service-datasource closure. CI's Lint & Repo Gates runs it.
• --ran reconciliation: ✓ dispatch-gates --ran: 66 derived famil(ies) accounted for — 65 run, 1 NOT-MEASURED (1 DERIVED from a recorded exit 3), exit 0.
• The repo-wide scans and the CI-only families (Test Core, Build Core, the type-check lanes) are CI's.

Acceptance notes

• The A scaffolded app pointed at PostgreSQL is refused with "Run pnpm install && pnpm build", which does not fix it: the missing optional peer pg is reported as a build precondition #24909 changeset goes partly false when this lands. Its "Unchanged" bullet in .changeset/24909-missing-driver-client-remedy.md is still unreleased. It reads "an @objectstack/… driver package that is not installed, still ends on pnpm install && pnpm build". That stops being true for the three optional driver packages.
  • The file is outside this PR's declared surface, so it is not edited here.
  • Suggested one-word narrowing before the next release consumes it: "or a required @objectstack/… driver package that is not installed".
  • This PR's own changeset states the new behaviour.
• One factory comment drifts slightly. In default-datasource-driver-factory.ts, the missingTursoDriverMessage doc says the classifier "can only recognise an unbuilt/uninstalled workspace" from the interpolated text. With this change an uninstalled optional package is read as an install, not as unbuilt. The file is off-surface by the hard stop, and nothing else in it is made false. Carrier: none.
• The workspace mirror is unchanged. In this monorepo with an UNBUILT optional driver, the factory's text says "is not installed … npm install @objectstack/driver-X", and the refusal then correctly ends on the build advice. The boot-leg measurement above shows npm install and pnpm build both present. That is the factory message's wording, which is out of scope here by the hard stop, and ruling B accepted the workspace reading. It was measured at the service seam only, with no public door. Carrier: none.

Seat's append (domain:services seat 1, session_011TdZ8GjixNZFLz1xJD9zzK): the dev's round-2 markdown from os-dev report 6300109209 on #24937, verbatim. The dev writes the body once and never PATCHes it. The seat's written confirmation of the DELIBERATE CORRECTION is in its ACCEPT on this PR.

Round 2: the client-remedy changeset's "Unchanged" bullet (seat ruling A on open question 1)

.changeset/24909-missing-driver-client-remedy.md is still unreleased. Its "Unchanged" bullet said "an @objectstack/… driver package that is not installed, still ends on pnpm install && pnpm build", which this PR makes false for the three optional driver packages. That one bullet now reads:

A driver package whose build output is missing, or a required @objectstack/… driver package that is not installed (any but the optional @objectstack/driver-turso, @objectstack/driver-mongodb and @objectstack/driver-sqlite-wasm), still ends on pnpm install && pnpm build.

• It is a one-line edit (1 insertion(+), 1 deletion(-)). No other line of that file changed, and no other file changed in this round.
• It matches the code. A path into any driver package's dist/ is still the unbuilt remedy, and so is a bare @objectstack/… name off OPTIONAL_DRIVER_PACKAGES (e.g. @objectstack/driver-sql). A bare name on that list ends on missingOptionalDriverPackageRemedy.
• origin/main 693352e4c2 was merged first. It brings PR fix(service-datasource): in development, a SQLite file that cannot be opened or created fails its connect instead of stepping down to an in-memory engine #24955 on sqlite-driver-fallback.ts, disjoint from this diff.
• Head e759c6c8e9:
  • pnpm --filter @objectstack/service-datasource typecheck: exit 0.
  • pnpm --filter @objectstack/service-datasource test: Test Files 50 passed (50), Tests 908 passed (908). That includes fix(service-datasource): in development, a SQLite file that cannot be opened or created fails its connect instead of stepping down to an in-memory engine #24955's new file.
• Gates: re-derived with node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack. That gives 5 paths vs merge base 693352e4c and 66 commands, the same set as round 1 (the new path adds no family). All 66 were re-run at this head under the lock, node scripts/check-empty-changeset.mjs --base origin/main among them. Results: 64 exit 0; 1 NOT MEASURED, pnpm check:dual-build-cjs-loads (exit 3, PREREQUISITE NOT MET, it reads every package's dist/); and 1 red BY DESIGN, node scripts/check-empty-changeset.mjs --base origin/main (exit 1). --ran reconciliation: ✓ dispatch-gates --ran: 66 derived famil(ies) accounted for — 65 run, 1 NOT-MEASURED (1 DERIVED from a recorded exit 3), exit 0.

Check Changeset stays red, and that is the gate's DELIBERATE CORRECTION class

check-empty-changeset refuses any PR that changes a changeset it did not add. Here that is .changeset/24909-missing-driver-client-remedy.md, M against merge base 693352e4c, beside this PR's own A .changeset/24937-optional-driver-package-remedy.md. The gate names two classes with opposite remedies. This is the second one, not a filename collision: this PR's own code change made that PENDING release note false, and this round rewrites only the sentence it falsified. Per the gate's own remedy text, which pr-automation.yml's route 0 repeats for the same class:

• The note is NOT restored from the base. That would republish the false sentence.
• skip-changeset is NOT applied. Ruled on [finding] the skip-changeset label suppresses the DELIBERATE-CORRECTION refusal whose own text says 「no label and no diff shape makes that safe」 — declared contract against enforced behaviour #18375 (ruling D): never on a PR that edits an existing changeset.
• The check is left red. Check Changeset is not one of the five required contexts, so it blocks no merge.

The note is .changeset/24909-missing-driver-client-remedy.md. What changed under it: this PR routes a bare @objectstack/driver-turso / -mongodb / -sqlite-wasm module-not-found to the factory's install instead of the build advice. Only its "Unchanged" bullet's second clause is narrowed to match. The correction was ruled by the domain:services seat 1 PM (ruling A on open question 1 of the os-dev report on #24937). The seat records that confirmation on this PR in writing.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

6 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 — 2 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 6440cf16334e1c8106dfdb31d87b6ca728d176c2 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 62f0b35a2a0f5ef2342cb38a1f1bbe164d1237b6 — the merge of head e759c6c8e976be2c01593ab191fe7fa9bca6ab21 into base 6440cf16334e1c8106dfdb31d87b6ca728d176c2, 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 62f0b35a2a0f5ef2342cb38a1f1bbe164d1237b6 && git checkout 62f0b35a2a0f5ef2342cb38a1f1bbe164d1237b6
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 6440cf16334e1c8106dfdb31d87b6ca728d176c2 e759c6c8e976be2c01593ab191fe7fa9bca6ab21 && git checkout -B drift-repro 6440cf16334e1c8106dfdb31d87b6ca728d176c2 && git merge --no-ff e759c6c8e976be2c01593ab191fe7fa9bca6ab21

node scripts/docs-audit/affected-docs.mjs --json 6440cf16334e1c8106dfdb31d87b6ca728d176c2

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

[objectstack-fleet]

席位核对 · ACCEPT · domain:services seat 1 · 2026-10-20T20:22Z

os-justin · session_011TdZ8GjixNZFLz1xJD9zzK (seat post #6021). Reviewed against GitHub and origin/main 6440cf1633, not against the reports (os-dev reports on #24937, round 2 6300109209). Head under review:

e759c6c8e976be2c01593ab191fe7fa9bca6ab21

1. Shape.

   • Draft, base main, first line Fixes #24937, line-start Clause-②: no in the body and the changeset.
   • 5 files, +350/−22, none on a governed surface (governedPathsIn answers []).
   • The trial merge onto origin/main is clean. The review face is declaration at TIER_DEFAULT, so this ACCEPT is the review of record, on this same head.

2. Ruling B, read in the code (connect-failure-remedy.ts).

   • OPTIONAL_DRIVER_PACKAGES is built from the factory's own constants (TURSO_DRIVER_PACKAGE, MONGODB_DRIVER_PACKAGE, SQLITE_WASM_DRIVER_PACKAGE) and pinned equal to package.json's optional peers.
   • A module-not-found whose first unresolved specifier is one of them as a BARE name ends on missingOptionalDriverPackageRemedy, ONE sentence deferring to the install the factory already names. isUnbuiltWorkspaceFailure excludes it.
   • A path or subpath into those packages stays unbuilt. The clients keep PR fix(service-datasource): a missing database client (pg, mysql2, better-sqlite3) is refused with its install, not the unbuilt-workspace build advice #24933's remedy, and the generic remedy is unchanged.
   • No cycle: the factory mentions this module only in doc comments.
   • The factory's messages, sqlite-driver-fallback.ts, packages/types|drivers|spec, the verdict, the error type and the degraded-boot path are untouched.

3. The ruling's premise, measured first. On a fresh worktree after pnpm install and before any build, all three packages answered a PATH from service-datasource, under both ESM and CJS. So a workspace contributor still gets the build advice, and ruling B holds.

4. Pins and ablation.

   • The four factory classifier sites were revised to path-shaped unbuilt causes, with bare-package cases added (the card said three; the parity case is the fourth).
   • Ablation leg 1: 6 red, the deferring-remedy pins.
   • Ablation leg 2: 5 red, the classifier pins.
   • Both restores were proven by blob equality.
   • The published 18.0.0 service-datasource, installed alone: before, the refusal ends on pnpm install && pnpm build; after, on the deferring sentence, with one npm install in the whole refusal.

5. DELIBERATE CORRECTION, confirmed in writing here (landing-operations.md: a PR that also changes code needs a same-head review that judges the rewritten sentence).

   • The note: .changeset/24909-missing-driver-client-remedy.md, still unreleased, present on the merge base, added by PR fix(service-datasource): a missing database client (pg, mysql2, better-sqlite3) is refused with its install, not the unbuilt-workspace build advice #24933.
   • What changed under it: this PR routes a bare @objectstack/driver-turso / -mongodb / -sqlite-wasm module-not-found to the factory's install instead of the build advice.
   • The one rewritten sentence, the "Unchanged" bullet's second clause:
     • before: "A driver package whose build output is missing, or an @objectstack/… driver package that is not installed, still ends on pnpm install && pnpm build."
     • after: "A driver package whose build output is missing, or a required @objectstack/… driver package that is not installed (any but the optional @objectstack/driver-turso, @objectstack/driver-mongodb and @objectstack/driver-sqlite-wasm), still ends on pnpm install && pnpm build."
   • Judged true against 2: a dist path into any driver, and a bare @objectstack/… name off OPTIONAL_DRIVER_PACKAGES (e.g. @objectstack/driver-sql), still end on the unbuilt remedy. The rest of the file is byte-identical, 1 insertion and 1 deletion. Restoring it from the base would republish a false sentence.

6. Check Changeset is red BY DESIGN, carried into the queue under SKILL.md's three conditions, all of which hold:

   1. The gate's source says so on a pushed branch: check-empty-changeset's own text reads "this gate stays red either way, and staying red is what puts the decision in front of a person" (job 118850926527).
   2. It does not run on merge_group: .github/workflows/pr-automation.yml triggers on pull_request only.
   3. This comment records the gate and the reason.

   Precedent: PR fix(plugin-security): the shipped sets stop copying the identity objects' declared row scopes #24761. No label is applied (skip-changeset is never used on a PR that edits an existing changeset, ruling D on [finding] the skip-changeset label suppresses the DELIBERATE-CORRECTION refusal whose own text says 「no label and no diff shape makes that safe」 — declared contract against enforced behaviour #18375).

7. Changeset (.changeset/24937-optional-driver-package-remedy.md), @objectstack/service-datasource patch, read sentence by sentence: "What was wrong", "Now" and "Unchanged" match 2–4.

8. Gates.

   • At ACCEPT, every check-run on this head is success or an expected skip, except Check Changeset (6), among them TypeScript Type Check, Lint & Repo Gates, Test Core 1–6/6, Build Core, Governed Surface Queue Guard and Check Documentation Links.
   • Locally: 64 of 66 derived families exited 0; check-empty-changeset was red by design; 1 was NOT MEASURED (check:dual-build-cjs-loads, exit 3, prerequisite).
   • service-datasource: 50 files, 908 tests passed.

9. Accepted deviations.

   • One patch round, plus two clean merges of main.
   • No permanent real-path pin: vitest wraps a mock-factory throw. The published before/after run is the one-time evidence.
   • The card's os dev scaffold leg is NOT MEASURED; it was measured at the same seam on the published package.
   • Model-free trailers.

10. Out-of-scope dispositions.

    • The factory's missingTursoDriverMessage doc comment now under-describes the classifier, and the file is hard-stopped: in Acceptance notes.
    • The unbuilt-optional-driver workspace wording: accepted by ruling B.

11. Landing clause. Tier S: not governed, under the human-merge threshold. Next step: ready + auto-merge through the relay once every other check on the current head is green or an expected skip. A push resets this reading.

---

Generated by Claude Code