docs(ci): two more sentences still say the merge queue runs the FULL suite — stale since #16453 (merge-queue-triage.yml, run-dev-unbuilt-workspace.e2e.test.ts) #16574

Found while correcting the two sites named by #16479. That card corrects AGENTS.md §7 and scripts/check-cross-package-test-inputs.mjs's header; a repo-wide grep for the same claim turns up two MORE present-tense sentences that #16453 (PR #16471) made false, in files the card does not name. Filed rather than folded in: #16479's dispatch scoped it to exactly two sites.

The true state on main today (scripts/ci/select-shard-packages.sh): merge_group runs the affected set against the group's own base_sha, unioned with the cross-package scans. Only push on main still runs the FULL package list.

Site 1 — .github/workflows/merge-queue-triage.yml:9-11

# A queue failure is a different animal from a PR failure: the PR's own CI ran
# affected-only, while the queue runs the FULL suite on the speculative merge
# result. The failing test is therefore often in a package the PR never
# touched — a flake, or a semantic conflict with another queued PR — and

The premise is now false. The conclusion still holds, but for a different reason: the queue's affected set is computed against the GROUP's base, so it covers every entry in the group, not just this PR's diff. This comment is the stated rationale for the whole workflow, so a stale premise here misinforms exactly the reader who is diagnosing a queue red.

Site 2 — packages/cli/test/run-dev-unbuilt-workspace.e2e.test.ts:49-51

 * that is the one reader for which this diagnostic was never at risk. The merge
 * queue is not that reader: it runs the full suite sharded, and a worker whose
 * loop is starved stops draining its children for seconds at a time.

Present tense, now false. The load argument the docblock builds on it (a starved event loop on a contended shard) survives — the queue still runs sharded, just over a smaller set — so this is a wording correction, not a change of conclusion.

⚠️ Not stale, deliberately: the same file's line ~182 (Merge-queue shards running the full suite six ways sharded went over it three times in a day) is PAST tense about a measurement that happened, and scripts/check-cross-package-test-inputs.mjs's Layer B is why the merge queue did not catch it is past tense about #7802. Both are correct history and should be left alone.

Scope

Text only, two files, no CI behaviour change. Neither file is a governed surface, so this one can go through the queue normally.

Generated by Claude Code