feat(qa): os test prints suite and scenario names and selects scenarios by --tags

[objectstack-fleet]

Part of #20289

Clause-②: no

Rework round 1

The seat's order of record is comment 5861125164 on #20289, which also amended the claim's file surface to these three files, text only. This PR had made three texts false or stale. Commit 5d31b4d587 corrects them. It is a new commit on top of dd6b472c63, with no rebase, amend or force-push.

1. docs/qa/platform-checklist/areas/cli.json, item cli.qa-suite-execution:
   • The knownGap that read "no scenario SELECTION exists: os test has exactly three flags …" now says what stays true. --tags selects scenarios (any-of, exact, case-sensitive). Deselected scenarios are counted and never passed. requires is declared but NOT CHECKED, pending the maintainer's decision on qa: os test prints suite and scenario names, filters by tags, and skips on unmet requires (5 keys) #20289.
   • The item's source note on qa.json#tags said the same false thing ("its dead tags/requires rows are why no scenario selection exists"), so it is corrected in the same edit.
   • revision goes from 4 to 5 with a history entry, as the checklist's change rule requires.
2. content/docs/deployment/cli.mdx, § os test:
   • Documents --tags: comma list, any-of, exact and case-sensitive. Deselected scenarios are counted and never passed. A listed tag no loaded scenario carries is named, and an empty entry is refused. A selection that matches no scenario exits 0, or 1 under --fail-on-empty.
   • Documents the report lines: the suite heading NAME (FILE), the scenario line NAME [ID], and a failed scenario's description.
   • States that requires is not checked.
   • Adds one example line. The page agrees with os test --help, read at this head.
3. packages/spec/scripts/liveness/check-liveness.mts, the qa paragraph of the header comment: suite.name and scenario.tags now have readers in os test. scenario.requires is the one key still unread: declared, NOT CHECKED, and its ledger row stays dead.

This round changes no code, and the requires question stays with the maintainer.

Verification at 5d31b4d587:

• dispatch-gates --repo objectstack-ai/objectstack --commands was re-derived for the widened diff. It now names 118 families: the 114 of round 0 plus check:cli-examples-parity, check:pm-dispatch-gates, check:pm-governed-merges and bare-root-worklist.mjs --self-test.
• The first pass ran while the fresh worktree's prerequisites were still being rebuilt. The gates that read built output answered exit 3 (or exit 1 for check:dts-closure, "declarations missing").
• A clean second pass ran every family except check:pm-dispatch-gates once the rebuild had finished. All 117 exited 0. check:pm-dispatch-gates reads no built output; it exited 0 on the first pass after 1270 s.
• Reconciliation with --ran (every line is COMMAND :: exit CODE): 118 derived, 118 run, 0 NOT-MEASURED, 0 UNRUN.
• check:platform-checklist: OK — 15 areas, 266 items.
• The edited script's own tests are the 8 files that name check-liveness, 6 of them beside it in scripts/liveness/. All passed: 8 files, 262 tests.

os test now reports the suite and scenario names an author writes and selects scenarios with --tags. That makes four of the qa-runner family's five ledger keys live. The fifth, TestScenario.requires, is not guessed. This PR leaves it dead with an honest describe(), and the report on #20289 puts the question to the maintainer. The card stays open for that half. This PR does not deliver the "unmet requires is SKIPPED with its reason" leg or its pin.

What changes

• @objectstack/core: QA.TestResult gains scenarioName and description on both result envelopes: the completed run and the setup-failure early return. It also gains suiteName on every result that runSuite produces. suiteName is absent only from a lone runScenario call, which has no suite.
• @objectstack/cli (os test):
  • The suite heading is 📄 Running suite: NAME (FILE). A file that fails to load is still headed by the file alone.
  • Each scenario line is Scenario: NAME [ID] (Nms), and shows the id once when the name and id are equal.
  • A failed scenario prints its description under its line.
  • A new --tags TAG[,TAG...] flag. Deselected scenarios are counted on the summary and never counted as passed. A requested tag that no loaded scenario carries is named there. An empty entry is refused at parse time.
• @objectstack/spec: only the describe() of requires, requires.params and requires.plugins changed. They now say NOT CHECKED instead of reading as a guard, and the reference page was regenerated. In liveness/qa.json, name, scenarios.name, scenarios.description and scenarios.tags move to live and cite their readers as file#symbol. tags also cites a producer, the --tags value. requires stays dead with a re-measured note. The state-counts.md row moved from live 4 / dead 5 to live 8 / dead 1, and the README row was rewritten to match.

--tags semantics: any-of, exact, case-sensitive

A comma list selects a scenario that carries at least one listed tag. Why this form:

• It is the reading of a comma list in Odoo's --test-tags, a metadata platform, and in Cucumber's comma form.
• It is the everyday use of Playwright's --grep "@smoke|@critical": name the sets you want and get their union.
• An all-of form or an expression language (Cucumber tag expressions, pytest -m) is a grammar to specify, parse and report errors against. No suite has needed one. The only suite in this repo, examples/app-showcase/qa/platform-smoke.test.json, tags one scenario smoke, crud.

Matching is exact and case-sensitive, because a tag is a name the author chose, not a pattern. With the flag, an untagged scenario is never selected. Without the flag, nothing changes: the selection is not consulted and every scenario runs. The help text states all of this.

Exit status, and the convention it follows

The repo's os test convention is that a run that executed nothing is not a failure by default, and --fail-on-empty is the opt-in strict reading. It is stated in Test.description in packages/cli/src/commands/test.ts and in content/docs/deployment/cli.mdx, § os test: "A pattern that matches no suite is not a failure by default".

A --tags selection that matches no scenario takes that same posture:

• It prints No scenario matched --tags X. and exits 0.
• Under --fail-on-empty it exits 1. The flag's description now names both cases.

A renamed tag in a CI step is the same trap as a renamed directory, and it gets the same opt-out. The Found N test suites., SUCCESS: All N scenarios passed. and FAILED: N scenarios failed. M passed. lines keep their spelling. The platform checklist item for os test reads the latter two.

Measured before and after

The target was a stub HTTP server that answers every request with 200. The suite had four scenarios:

• one tagged smoke;
• one tagged regression;
• one declaring requires: { plugins: ["plugin-that-does-not-exist"], params: ["OS_TEST_PARAM_THAT_IS_NOT_SET"] };
• one plain control.

probe	base 10ea9eb2ed	this branch
suite heading	📄 Running suite: probe.test.json	📄 Running suite: Probe suite: names, tags and requires (probe.test.json)
scenario line	✅ Scenario: s-tagged-smoke (28ms): the id only (control: the id is printed)	✅ Scenario: A scenario tagged smoke [s-tagged-smoke] (23ms)
--tags smoke	exit 2, Nonexistent flag: --tags (control: --fail-on-empty parses, exit 0)	exit 0; 1 scenario run; --tags smoke selected 1 of 4 scenarios; 3 deselected (not run, not counted as passed).
unmet requires	✅ PASSED, same as the no-requirements control	unchanged, deliberately: see the next section

requires: why it is left dead and put to a decision

The dispatch rule was: "A sub-key you cannot judge honestly is a needs_decision, and ⛔ you don't guess." Neither sub-key can be judged honestly today:

• plugins: os test reaches the target only over HTTP, through HttpTestAdapter. No server surface lists the loaded plugins. The discovery document (DiscoverySchema) advertises services and capabilities, not plugins. /.well-known/objectstack does not list them either. kernel.getPlugins() is in-process only. The spelling is also undefined: the npm package name or plugin.name.
• params: it is undefined whose environment is meant. The runner's process.env is observable, but no step can consume a runner variable, because resolveVariables interpolates only captured context. The server's environment is not observable at all.

The decision is posted on #20289 with its options and the four-axis analysis. Until it is taken, the describe() says NOT CHECKED rather than implying a guard. The PR contains no speculative SKIPPED machinery, because every enforce option needs it and the retire option would leave it dead.

Verification

Every run below is at dd6b472c63, the final commit.

• core:
  • pnpm --filter @objectstack/core test: 56 files and 1483 tests passed. src/qa/runner.test.ts has 20 tests, 3 of them new: names on every result, names on the setup-failure envelope, and a lone runScenario with no suite.
  • pnpm --filter @objectstack/core typecheck: exit 0.
• cli:
  • vitest run --project unit: 227 of 230 files passed on the first run. Two files failed because packages/cli was not built yet. One file, hook-timeout-override-refusal.test.ts, timed out twice at 5000 ms under an 861 s shared-box run. After the CLI build, those 3 files passed 33 of 33 on a re-run.
  • The new test/qa-tags-selection.test.ts (unit) has 13 tests, including the unfiltered CONTROL.
  • The new test/qa-names-and-tags-run.test.ts (integration: it spawns bin/run-dev.js against a node:http stub) has 8 tests. They pin the names printed, --tags smoke exiting 0 because the failing scenario never ran, the unfiltered control exiting 1, a no-match selection exiting 0 by default and 1 under --fail-on-empty, and a malformed list refused before anything runs.
  • pnpm --filter @objectstack/cli typecheck: exit 0.
• spec:
  • vitest run src/qa: 26 tests passed.
  • check:generated: check:docs was the one stale artifact. --fix regenerated it, and it is clean on re-check.
  • check:liveness: green, with qa 9 classified (live 8, dead 1).
• Ablation (one leg, node scripts/ablation-replace.mjs, WRAP mode):
  • The mutation was return hits.length > 0; → return true; in selectScenariosByTags. The tool reported anchor 1 -> 0, and the blob went from 774d2436cf4f to cab531acb335. That is the only mutation.
  • Unit: 3 red out of 13. They were the any-of, exact-match and no-match pins. The CONTROL stayed green.
  • Integration: 3 red out of 8. They were --tags smoke, --tags regression and no-match. The unfiltered control and the names pins stayed green.
  • The direction was the expected one: the tests went red.
  • Restore: the blob after restore equals HEAD (774d2436cf4f), and git diff HEAD is empty.
  • The subject is imported through a relative path to src/, so no dist/ leg applies.
• Gates: there are 114 derived commands (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --commands).
  • 108 exited 0 on the first pass.
  • 6 answered PREREQUISITE NOT MET (exit 3), which means not measured. After the missing builds, they were re-run: all 6 exited 0. They were check:skill-examples, check:dual-build-cjs-loads, check:i18n, check:i18n-coverage, check:i18n-walk-parity and check:type-check-debt (the last reported --re-measure: OK … none above its recorded number). The first pass of check:type-check-debt had died in its own internal build: plugin-pinyin-search could not find @objectstack/objectql declarations while a prerequisite build was rewriting dist/. The gate labelled that pass not measured itself.
  • Reconciliation with --ran (every line is COMMAND :: exit CODE): 114 derived, 114 run, 0 NOT-MEASURED, 0 UNRUN.
• Lint, measured narrowly:
  • pnpm exec eslint --no-inline-config --format json over the 6 changed TypeScript files: 6 results, 0 errors, 0 warnings.
  • The population comes from eslint's own config: isPathIgnored is false for all 6, and the 5 changed JSON, Markdown and MDX files are ignored.
  • Invariance: this repo's config enables no type-aware linting. parserOptions.project and projectService resolve to null for all 6, so this diff cannot move any untouched file's verdict.

Acceptance notes

• The three texts that were outside the claimed file surface in round 0 (cli.mdx, the checklist knownGap and the checker's comment) are corrected in Rework round 1 above, after the seat amended the surface in 5861125164.
• Semver:
  • @objectstack/cli is minor, for the new flag.
  • @objectstack/core is also minor, not patch. QA.TestResult gains three public fields, which is an additive API that programmatic consumers can read.
  • @objectstack/spec is patch, because only describe text and ledger data changed.
  • All three are in the same changesets fixed group.
• Branch base: origin/main at d498113b50 is merged in with scripts/pm/os-regen-merge.sh, as merge commit 9056455134. The one conflict was packages/spec/liveness/README.md: main's translation and rest_api rows and this PR's qa row are all kept. The generated state-counts.md was regenerated, not hand-resolved, in 76e9a48821: qa is 8/1 on top of main's totals. Against that merge, git diff d498113b50 HEAD touches only this PR's 14 files.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 3 package(s): @objectstack/cli, @objectstack/core, @objectstack/spec, touching 16 documentable anchor(s). ⚠️ 3 changed file(s) yielded no anchor (packages/spec/liveness/README.md, packages/spec/liveness/qa.json, packages/spec/liveness/state-counts.md), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

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

• content/docs/deployment/cli.mdx (via os test (command, read off packages/cli/src/commands/test.ts))
• content/docs/protocol/kernel/lifecycle.mdx (via os test (command, read off packages/cli/src/commands/test.ts))

What this run could not see

• 3 changed file(s) yielded no anchor (packages/spec/liveness/README.md, packages/spec/liveness/qa.json, packages/spec/liveness/state-counts.md) — pages documenting those are invisible to this run
• 4 name(s) were too generic to anchor anything (single lowercase words)
• 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 — 146 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 d498113b505f7b29b5e2c554f3152145edeb3461 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 51b76c46ea20de097e32027226e3e8f292d09060 — the merge of head 5b15c28fb146529543f02290ce18bf81d0f63df8 into base d498113b505f7b29b5e2c554f3152145edeb3461, 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 51b76c46ea20de097e32027226e3e8f292d09060 && git checkout 51b76c46ea20de097e32027226e3e8f292d09060
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin d498113b505f7b29b5e2c554f3152145edeb3461 5b15c28fb146529543f02290ce18bf81d0f63df8 && git checkout -B drift-repro d498113b505f7b29b5e2c554f3152145edeb3461 && git merge --no-ff 5b15c28fb146529543f02290ce18bf81d0f63df8

node scripts/docs-audit/affected-docs.mjs --json d498113b505f7b29b5e2c554f3152145edeb3461

⚠️ 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 d498113b505f7b29b5e2c554f3152145edeb3461 → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

Contract review

Served-tier: 68/68 CONTRACT_REVIEW_TIER
Head-sha: 5d31b4d587c922595840651596b8e8125116dbd6

① Derived judgments

1. Four keys live, readers real. qa.name: runner.ts#runSuite spreads suiteName: suite.name onto every suite result; test.ts#suiteHeading prints 📄 Running suite: NAME (basename). scenarios.name/description: runner.ts#runScenario stamps scenarioName/description on BOTH envelopes (setup-failure return and completed run); test.ts#scenarioLabel prints NAME [ID], id once when equal; description printed only under a ❌ line (test.ts#run). scenarios.tags: test.ts#selectScenariosByTags, producer #parseTagsFlag. Every path#symbol anchor resolves under evidence.mts#isSymbolNamed; Spec property liveness green. Nit: README row says the name "rides every TestResult"; a lone runScenario result has no suiteName (qa.json's "every TestResult the suite produces" is the exact wording).
2. --tags: any-of/exact/case-sensitive is tags.includes(tag) over scenario.tags ?? []; comma list trimmed, deduped; an empty entry throws inside the flag parse, before any suite loads. Deselected scenarios never reach runner.runSuite, so totalPassed cannot count them; they are counted on tagSelectionLine. No-match branch tags && totalSelected === 0 exits 0, or 1 under --fail-on-empty, and is reached only when totalFailed === 0: a load-refused or throwing suite still takes FAILED/exit 1. Same posture as the pre-existing Found 0 test suites. branch and Test.description (os test: 5 of the 8 declared action types cannot reach a stock server (wrong base path), and a zero-match glob exits 0 #7848). Pinned: qa-tags-selection.test.ts (13, incl. CONTROL) and the spawn pin qa-names-and-tags-run.test.ts (8: control exits 1; --tags smoke exits 0 with the failing id absent; no-match 0/1; malformed non-zero with no suite run). cli test is vitest run over all tiers, Test Core shards green. No path reads a filtered-out or erroring scenario as passed. Non-blocking: (a) the unknown-tag warning is gated on a positive totalSelected, so --tags X matching nothing beside a load-refused suite prints only --tags X selected 0 of N; (b) a malformed list exits 1 via oclif handle() (err.oclif?.exit ?? 1; parse.js rethrows the plain Error; no exitCodes.failedFlagParsing configured), not the CLIParseError exit-2 + INVOCATION ERROR line an unknown flag gets (invocation.ts#isInvocationError); the pin asserts non-zero only.
3. TestResult gains suiteName?, scenarioName (required), description?: additive for readers. Grep at head (objectstack + objectui): the only consumer is scenarioLabel via Pick; no producer outside runner.ts; TestExecutionAdapter untouched; check:api-surface green. Printed lines match the docs. Three new core pins cover both envelopes and the lone runScenario.
4. requires: the zod hunk is describe text plus one comment; row stays dead, verifiedAt 2026-09-27, note records the stub re-measure. No text in the diff claims a check (changeset, cli.mdx, cli.json, check-liveness.mts, README, generated mdx all say NOT CHECKED / nothing checks). The describe ("NOT CHECKED by os test or the core TestRunner: the scenario runs whether or not they hold…") is unambiguous. Nit: cli.mdx, cli.json and the changeset say a scenario naming a missing plugin "still runs, and fails on whatever the missing plugin causes"; it can also pass (the PR's own probe did). The describe's "surfaces only as the failure it causes" is the accurate form.
5. cli.json: knownGap and the qa.json#tags source note are true at head; revision 4→5 with a history entry; steps/acceptance unchanged and clause 1 ("✅ with its scenarioId") still holds (fixture names ≠ ids, so the id is bracketed). Nit: the history says the id is "now in brackets after the name"; it is bare when equal. cli.mdx: every sentence matches the code and agrees with Test.description and the --tags/--fail-on-empty flag text; quoted lines equal tagSelectionLine and No scenario matched --tags …. check-liveness.mts: comment only, true.
6. testing.mdx carries the describe strings verbatim; check:docs runs at lint.yml:5564 (Lint & Repo Gates green). state-counts.md qa 8/1, total 937/164/1119: hand recount of qa.json agrees; check:liveness byte-checks the artifact (countsArtifactErrors), green. Exact at the PR tree; stale once main is merged (③).
7. Changeset: every sentence true (nit in 4). cli minor ✓ (new flag). core minor ✓ (three public fields on exported QA.TestResult; the required scenarioName would break only a producer, and none exists). spec patch ✓ (no shape change). All three sit in one fixed group (69 packages). Clause-②: no is its own line in the changeset and the PR body; readClause2Line reads declared/no/no arm. The zod diff widens no accept set.
   CI at head: 35 check-runs, 33 success, 2 skipped (Console Pin Gate, opt-in tarball smoke), 0 failed, run 00:15–00:38Z.

② Semver level

minor. @objectstack/cli minor (new --tags), @objectstack/core minor (additive public fields), @objectstack/spec patch (describe text and ledger data; check:api-surface green); one fixed group ⇒ minor release. Clause-②: no with no arm per scripts/pm/clause2-line.mjs: nothing widens or narrows an accept set, so AGENTS.md's "yes takes at least minor / (narrowing) is BREAKING" rules do not bite. Check Changeset green.

③ Boundary flags

• requires: triage 5859588706 ENFORCE covers all five keys, including "unmet requires reported skipped with its reason, never passed". This PR delivers four, leaves the row dead with an honest describe, and opens Part of #20289; the maintainer's word is owed on the card, not in this PR.
• git merge-tree --write-tree origin/main origin/claude/issue-20289-os-test-names-tags-requires (main = 26daf0b) is NOT clean: CONFLICT (content) in packages/spec/liveness/README.md (main 826f327 at 00:23Z edited the translation row :929 and 26daf0b at 01:26Z the rest_api row :939; the PR edits the adjacent qa row :930), and state-counts.md is "generated, not text-merged" (main total 936/166/9/1117 vs PR 937/164/12/1119). GitHub mergeable_state: dirty; CI ran before both commits. Fix: merge main, keep both README rows, run pnpm --filter @objectstack/spec gen:liveness-counts.

Implemented-by: claude/issue-20289-os-test-names-tags-requires
Reviewed-by: session_01QcAS3qiYYZNezaxZxaUdMV

VERDICT: PASS

[objectstack-fleet]

Contract review

Served-tier: 94/94 CONTRACT_REVIEW_TIER
Head-sha: 5b15c28fb146529543f02290ce18bf81d0f63df8

① Derived judgments

1. Diff vs the new base d498113b50: the same 14 files, 624+/42−. A -U0 diff-of-diffs (hunk headers stripped) against the reviewed 10ea9eb2ed..5d31b4d587 differs only in: the changeset spec bullet (nit a); cli.mdx :1702–1703 (nit a); cli.json knownGap :456 (nit a) and history :561 (nit b); the README qa row :930 (nits a, c); and the state-counts.md total row, because the base moved 933/168/12/1119 → 936/166/9/1117. No other difference. runner.ts, test.ts, testing.zod.ts, qa.json and check-liveness.mts blobs are identical to 5d31b4d587. Merge 9056455134 (parents 5d31b4d587, d498113b50) adds no file beyond the 14; commit 76e9a48821 touches only state-counts.md; 5b15c28fb1 touches the four text files, 6/6 lines.
2. README: vs d498113b50 only the qa row differs (1 insertion, 1 deletion); main's translation and rest_api rows hash-equal between d498113b50 and the merged tree. state-counts.md vs base: only qa 4/5 → 8/1 and total 936/166 → 940/162, other columns unchanged; arithmetic exact. check:liveness byte-checks the artifact (countsArtifactErrors); Spec property liveness is green at head.
3. Nit texts, each TRUE at head: (a) "still runs, and the unmet requirement surfaces only as whatever failure it causes, if any" (cli.mdx adds "the scenario can still pass"; cli.json and README carry the same clause). Grep of the merged tree: nothing in packages/core/src/qa or commands/test.ts reads scenario.requires. (b) history: "now in brackets after the name, or bare when the name equals the id" matches scenarioLabel. (c) README: "stamped as suiteName on every TestResult the suite produces" matches runSuite; a lone runScenario is no longer over-claimed.
4. Main 10ea9eb2ed..d498113b50 touched none of the qa surfaces except cli.mdx, in the os g sections (hunks :1406–1460, :2168–2171), disjoint from the PR's os test hunks (:1654–1716), auto-merged. Main's diff mentions no TestResult, os test, --tags, scenario.tags/requires, TestSuiteSchema or liveness/qa.json. In the merged tree the sole TestResult consumer is still scenarioLabel via Pick, and there is no requires reader.
5. CI at head: 42 check-runs, 38 success, 4 skipped (Console Pin Gate, opt-in tarball smoke, and the 03:24Z re-fire of Auto Label / Check PR Size after the body edit), 0 failed. Lint & Repo Gates, Spec property liveness, Test Core and its 6 shards, Check Changeset, the five Type Check jobs and Build Docs are green. PR body's branch-base bullet is now true; mergeable_state: clean.

② Semver level

Unchanged: minor. cli minor, core minor, spec patch, one fixed group; Clause-②: no, no arm; the delta is docs text and a regenerated count.

③ Boundary flags

• merge-tree vs current origin/main 29720975b6 (5 commits past d498113b50): driver-free, from a shared bare clone with no merge.os-regen driver, writes tree 3f9cc93755, exit 0, no conflict. The checkout's driver run also writes a tree and only notes that content/docs/references/qa/testing.mdx is generated: main's 9daced0b6f rewrote its frontmatter description: after d498113b50, so the next main merge owes gen:schema + gen:docs; the PR's copy is exact for its own tree (check:docs green).
• requires: still the maintainer's decision on qa: os test prints suite and scenario names, filters by tags, and skips on unmet requires (5 keys) #20289; this PR leaves the row dead with an honest describe and opens Part of #20289.

Implemented-by: claude/issue-20289-os-test-names-tags-requires
Reviewed-by: session_01QcAS3qiYYZNezaxZxaUdMV

VERDICT: PASS