Skip to content

feat(queen): the review valve -- no person in the loop (Closes #6616) - #6617

Merged
gHashTag merged 2 commits into
masterfrom
queen-review-valve
Oct 5, 2026
Merged

gHashTag merged 2 commits into
masterfrom
queen-review-valve

Conversation

@gHashTag

@gHashTag gHashTag commented Oct 5, 2026

Copy link
Copy Markdown
Owner

Closes #6616

What

specs/queen/review_valve.t27 is the one home of the owner's rule of 2026-10-05: no person is needed in the review loop. Every review outcome that is not an accept gets an automatic next step and a clock:

outcome next step clock
empty diff, reviewer gave up, dead letter, unrecorded reason release as failed (retry) 30 min (RETRY_FLOOR_MINUTES)
send-back ceiling release once 60 min (CEILING_FLOOR_MINUTES)
no criteria, criteria beyond the patch, accepted on base truth alone criteria backfill; the rewrite releases the row close after 120 min (BACKFILL_WINDOW_MINUTES)
releases spent (MAX_RELEASES = 1) close as obsolete, reason recorded, boundary freed at once
reviewer bee REQUEST_CHANGES on a queen-* PR send-back to the Queen on the same branch at most PR_FIX_ATTEMPTS = 2, then the PR is closed (also after 48 h with no push)

The test no_kind_waits_for_a_person checks that, at the policy's own 48-hour hold, no kind of escalation still holds its files.

Why these numbers

Measured 2026-10-05: 144 of 219 review cards were escalated. Of the 193 stuck branches, 125 were empty. fileConflict was 108, and 25 of 70 workers were active. The constants reuse the supervisor's measured floors (EMPTY_ATTEMPT_FLOOR_MS, CEILING_RELEASE_MS, maximumSendBacks). There is one release per issue because 65% of the stuck branches are empty after three silent attempts each. The reasoning for each constant is in its doc comment.

Files

  • specs/queen/review_valve.t27: the spec, with 8 tests and 1 invariant
  • gen/c/queen/review_valve.c: t27c gen-c output. cc -DT27_TEST_MAIN gives All 8 tests passed., and a mutant that never releases fails test_an_empty_attempt_is_released_after_the_floor
  • docs/now/2026-10-06-queen-review-valve.md

Not here: the gHashTag/BrowserOS wiring (a separate PR; the owner merges it). The tools/bees/reviewer.py and criteria_backfill.py changes are described in #6616 and not written.

🤖 Generated with Claude Code

…step and a clock (Closes #6616)

specs/queen/review_valve.t27 is the one home of the owner's rule of
2026-10-05 ("no person is needed"): escalations release as failed after a
floor (30 min; 60 min for a spent send-back ceiling), at most MAX_RELEASES=1
per issue, then close as obsolete with the reason recorded and the boundary
freed; missing or unrunnable criteria wait at most 120 min for a backfill;
a reviewer-bee REQUEST_CHANGES on a queen-* PR goes back to the Queen at most
PR_FIX_ATTEMPTS=2 times, then the PR closes (48 h without a push).

gen/c/queen/review_valve.c is t27c gen-c output: 8 tests pass with
-DT27_TEST_MAIN, a mutant that never releases fails.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Oct 5, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-05 21:32:07 UTC

Summary

Status Count
Total Open PRs 50
PRs with Failing Checks 38
PRs with All Checks Green 12
READY 9
FAILING 38
PENDING 0
NO CHECKS YET 0

These columns do not partition: 9 + 38 + 0 + 0 = 47, and there are 50 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=ab873301be9c != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@github-actions

github-actions Bot commented Oct 5, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-05 21:45:41 UTC

Summary

Status Count
Total Open PRs 50
PRs with Failing Checks 38
PRs with All Checks Green 12
READY 10
FAILING 38
PENDING 0
NO CHECKS YET 0

These columns do not partition: 10 + 38 + 0 + 0 = 48, and there are 50 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=ab873301be9c != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@gHashTag
gHashTag merged commit 98ead51 into master Oct 5, 2026
25 of 27 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Queen review valve: no person in the loop

1 participant