[objectstack-fleet]

Requested by Steve · project thread
Fixes #26092

This is PR 2 of 2, the vitest half. PR 1, the tsconfig half, was #26203 and is merged.

Clause-②: no
Metadata-protocol: none

What changes

• vitest.preset.mts (new, repo root) holds the one setting the 18 byte-identical package configs carried, test.disableConsoleIntercept: true, with its rationale once. It is a plain object, not defineConfig(...): the root package declares no vitest dependency, so importing vitest/config there resolves only through pnpm hoisting and adds a TS2307 to the root program (root DEBT measured 26 to 27 with the import, 26 without).
• The 18 configs (former md5 0d3c266a...) become a copyright line, one comment line and export { default } from 'REL/vitest.preset.mts';.
• scripts/check-console-intercept-disarm.mjs. The card's premise that the gate "reads the effective config; it goes green by construction" does not hold: the gate greps each config's comment- and literal-masked source for the setting, and with the preset alone it went red on all 18 (measured by the handoff thread). The gate is kept and learns one narrow shape: a package-root config whose only statement is a bare re-export of the root vitest.preset.mts is judged by the preset's source, with the same disarm, re-arm and projects checks, and the finding names the re-export. A config that imports the preset and wraps or merges it is not recognised and must still carry the setting itself. Five self-test cases added (battery floor 11 to 16).
• turbo.json: the test task's inputs gain $TURBO_ROOT$/vitest.preset.mts, so an edit to the preset re-runs the re-exporting suites instead of replaying cache.
• scripts/check-cross-package-test-inputs.mjs: TURBO_ROOT_FILE_INPUTS gains vitest.preset.mts (its self-test pins the set of repo-root files turbo.json declares as inputs and went red on the new input without it).
• .github/workflows/ci.yml: the core filter lists vitest.preset.mts beside the two tsconfig bases, so a diff confined to the preset still starts Test Core. node scripts/check-ci-filter-parity.mjs and its --self-test both exit 0 with the line (it is not judged a leftover).

Verification (head c382986, on main 81bba6c)

• node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 107 commands on this head; all 107 ran, each exit captured before any pipe: 107 exit 0. --ran reconciliation (record carries every exit code): 107 derived, 107 run, 0 NOT-MEASURED, 0 UNRUN. Prerequisites first: clone unshallowed, every non-ui package built (76/76 tasks).
• node scripts/check-console-intercept-disarm.mjs --self-test: self-test OK: 15 cases + real-tree population floor (the five new cases among them); the scan: OK: 77 vitest-running package(s), every one disarms console interception at the package root.
• node scripts/check-ci-filter-parity.mjs and --self-test: exit 0.
• Package tests, turbo run test --force over all 18 re-exporting packages: 81/81 turbo tasks successful (dependency builds included); all 301 test files passed (driver-turso: 95 files, 34 tests skipped by its own environment guards). The same 18 also passed with this patch on main df8aac41f2 before the rebase. The suites were not re-run without the patch: every one passes with it, so none moved from pass to fail.
• Acceptance md5: the card's command as written (md5sum ... | sort | uniq -c) counts whole lines, hash plus path, so it prints 1 for any tree. Counting hashes only (awk '{print $1}' before sort | uniq -c), the top count is 13: identical copies of the one-line preset re-export, which the card allows. The other 5 re-exporters differ only in the relative depth. The remaining duplicate pairs predate this card (the create-objectstack blank template's src/dist copy; knowledge-memory/knowledge-ragflow).
• Gate ablation (handoff thread, same patch): removing the preset's disableConsoleIntercept: true reds the gate on all 18 re-exporters (exit 1); restored (hash-checked), exit 0.

Landing

Tier H: .github/workflows/ci.yml is a governed surface. This PR stays a draft until a GOVERNED_APPROVERS approval; the owning seat lands it after that.

No changeset (skip-changeset): the diff touches package vitest configs, a root config file, turbo.json, ci.yml and repo-internal gate scripts; no package's files[] ships a vitest config.

Acceptance notes

• A package with unit tests and no vitest.config.ts lets a stale @objectstack/core dist decide its verdicts — #7668 fixed one, nothing stops the next #7849 (a package with tests and no vitest config tests a stale dist): the preset makes the fix a one-line config. A package with unit tests and no vitest.config.ts lets a stale @objectstack/core dist decide its verdicts — #7668 fixed one, nothing stops the next #7849 remains open; this PR does not address it.
• The header line for check 2 of check-console-intercept-disarm.mjs still reads "its comment-masked source must set..."; the preset route is documented at the new constant only. A one-sentence header update could follow.

维护者速读(草稿)

• 改了什么:18 个内容完全相同的包级 vitest 配置改成一行转出仓根的 vitest.preset.mts(只含 disableConsoleIntercept: true);对应门禁学会认这种转出;turbo 测试输入和 ci.yml 的 core 过滤各加一行预设文件。
• 为什么改:同一设置抄 18 份，改一处要改 18 处;新包照抄也容易漏。卡 tooling: the root tsconfig.json is both the shared base for 72 packages and the scripts/ program (the one DEBT entry, 26 errors); 14 packages re-set module resolution three ways; 18 vitest configs are byte-identical; three debt ledgers are empty #26092 的第二个 PR。
• 风险与代价(含回滚):ci.yml 只加一行过滤条目，只会让更多 PR 触发 Test Core,不会少跑。回滚就是 revert 本 PR。
• 席位意见:
• 你要做的:因为碰了 .github/workflows/ci.yml,需要你批准(APPROVED review)后席位才能入队。

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

⚠️ 16 changed file(s) yielded no anchor (packages/connectors/connector-imap/vitest.config.ts, packages/connectors/connector-mcp/vitest.config.ts, packages/connectors/connector-rest/vitest.config.ts, …), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files. Nothing else in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 16 changed package(s)).

What this run could not see

• 16 changed file(s) yielded no anchor (packages/connectors/connector-imap/vitest.config.ts, packages/connectors/connector-mcp/vitest.config.ts, packages/connectors/connector-rest/vitest.config.ts, …) — pages documenting those are invisible to this run
• 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 — 40 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 0453fabef70c1b2a8276508d37c350d3487badce → packageMentionDocs.

[objectstack-fleet]

席位核对 · ACCEPT · devx seat 2 · 2026-10-26T18:05Z

Seat domain:devx#2 (#20163) · session_01Q4MDXRbEJCNdPAP4SiXwp9 · reviewed head c3829861f6 · card #26092, claim 6421431330 (delta M, amended). This is PR 2; PR 1 #26203 merged at 17:10Z. PR carries Fixes #26092.

• Scope: 23 files, +148/−317. ⛔ Tier H, because of one line in .github/workflows/ci.yml (the core filter gains 'vitest.preset.mts'). This seat ACCEPT does not lift the gate: the PR stays a draft until a GOVERNED_APPROVERS approval, and the seat will neither ready nor queue it before that. Nothing under ui/.
• The preset: vitest.preset.mts carries exactly test.disableConsoleIntercept: true with the finding: any late console.* in a vitest worker can fail a fully green suite — the #9371 amplifier is upstream and still live #10374 rationale. It is a plain object, not defineConfig, so the root program type-checks without a root vitest dependency.
• The 18 configs, by hash on this head: 13 are byte-identical and 5 more differ only in relative depth. Each is the one-line export { default } from '<rel>/vitest.preset.mts'. Configs that carry their own options (e.g. packages/core) are untouched, as the card asked.
• The cache inputs move together. turbo.json test.inputs gains $TURBO_ROOT$/vitest.preset.mts, and check-cross-package-test-inputs pins it in TURBO_ROOT_FILE_INPUTS. The ci.yml line means a preset-only diff starts the core jobs, and check-ci-filter-parity is green with it, so it is not a leftover.
• check-console-intercept-disarm learns exactly one re-export shape and judges the preset in its place. A config that wraps or merges the preset must still carry the setting itself. 15 self-test cases pass, including the 5 new ones.
• Dev evidence: 107 derived gates exit 0. On the 18 affected packages, turbo test --force ran 81/81 tasks and 301 test files, all passing.
• Note on the card's md5 acceptance command: it de-duplicates on hash + path, so it always prints 1. The seat counted by hash instead (above).
• Landing: after a GOVERNED_APPROVERS approval on this head, and once every check is green or skipped, the seat runs ready + auto-merge.

---

Generated by Claude Code