Skip to content

merge queue: checking #1324 + #1331 + #1334 + #1335 together on main (0d8f2e0) - #1338

Closed
mergify[bot] wants to merge 16 commits into
mainfrom
mergify/merge-queue/924c40fec7
Closed

merge queue: checking #1324 + #1331 + #1334 + #1335 together on main (0d8f2e0)#1338
mergify[bot] wants to merge 16 commits into
mainfrom
mergify/merge-queue/924c40fec7

Conversation

@mergify

@mergify mergify Bot commented Oct 11, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

🎉 This combination of pull requests has been checked successfully and will be merged soon. 🎉

#1324 + #1331 + #1334 + #1335 are queued together for merge on branch main (0d8f2e0).

This pull request has been created by Mergify to check the mergeability of [#1324 + #1331 + #1334 + #1335].
You don't need to do anything. Mergify will close this pull request automatically when it is complete.

Required conditions of queue rule admin-bypass for merge:

  • check-success = lint
  • check-success = test
  • check-success = validate

Required conditions to stay in the queue:

---
all_scopes: false
checking_base_sha: 0d8f2e02985d9359ca92df6b001426b2d7e836b1
previous_check_retries: []
previous_failed_batches: []
pull_requests:
  - number: 1324
    scopes: []
  - number: 1331
    scopes: []
  - number: 1334
    scopes: []
  - number: 1335
    scopes: []
scopes: []
...

EdbertChan and others added 16 commits October 10, 2026 23:28
@EdbertChan @cursoragent
Route restated existing rules to already-covered, not automate-me.
d55f636 
Reflect was writing a new preference whenever someone reminded the agent of a skill it had already skipped. Add lesson_route.py so that signal stays an intervention for ordering, and the lesson bucket is already-covered unless no skill states the preference yet.

Co-authored-by: Cursor <cursoragent@cursor.com>
Change-Id: Ibb2d3e90a7cf01987aeb14b71e9413d7cec536c3
@EdbertChan @cursoragent
Drop the lesson-router comments that failed the comment gate.
90c66e2 
The boundary-parser allow mark stays, and the comment checker treats that mark as a machine directive so the two gates agree.

Co-authored-by: Cursor <cursoragent@cursor.com>
Change-Id: I0a048d4d387df53de81d2638521a5198de499b36
@EdbertChan
Stamp an event_uid on every runner metrics row.
083c513 
Hook runs belonging to the same harness event had no shared key, so
grouping them for herd/wall-clock metrics relied on fragile 2s-window
inference. Add event_uid = sha256(stdin)[:12] to _row() so every
runner that receives the same stdin self-groups with zero coordination.
Invoker Bot
invoker: wf-1791728211949-11/publish-event-uid — Review claim: Every …
1e585e3 
…runner metrics row gains an event_uid field that groups all hook runs belonging to one harness event, with no change to hook stdout, stderr, or exit codes.

Review lane: behavior
Safety invariant: Hook stdout and exit codes stay the same. This slice does not change the other four claims.
Files: engine/hooks/_runner/run.py
Change types: bugfix
Acceptance criteria: The branch contains only this claim, and the review step publishes one pull request.

Exit code: 0
Invoker-Finalize-Id: 5330cf4f-7efd-43e4-9fc5-fbc19b9c9a89
@EdbertChan
Merge experiment/wf-1791728211949-11/publish-event-uid/g0.t0.a-a839f4…
4072085 
…770-8d121c9e — Review claim: Every runner metrics row gains an event_uid field that groups all hook runs belonging to one harness event, with no change to hook stdout, stderr, or exit codes.

Review lane: behavior
Safety invariant: Hook stdout and exit codes stay the same. This slice does not change the other four claims.
Files: engine/hooks/_runner/run.py
Change types: bugfix
Acceptance criteria: The branch contains only this claim, and the review step publishes one pull request.
CI Bot @claude
Print a before/after hook scorecard from the metrics ledger
af5796b 
`report.py --scorecard` folds runs.jsonl into one row per
(phase, event type) and prints procs/event, cpu_ms/event,
wall_ms/event, timeout%, crash% and spoke-rate. Rows are split into
"before" and "after" by whether the row carries a `dispatch_path`,
and grouped into events by `event_uid` (falling back to the row index
when a row has no uid), so one harness event counts once no matter how
many detectors ran under it.

The phase split and the proc count read the row field the runner
actually writes, `dispatch_path`. An earlier draft of this fold read
`dispatch_mode`, which no row has: every post-dispatcher row then
landed in the "before" bucket and the after-phase proc count never
credited the dispatcher, so the scorecard showed no change at all.

`--days N` is accepted alongside `--since` so the canary window can be
given in whole days.

Tests: two cases in engine/hooks/_runner/tests/test_report.py drive the
CLI over a fixture ledger holding one pre- and one post-dispatcher
event; they assert the per-phase numbers and the printed column header.
Both fail against the base report.py with "unrecognized arguments:
--scorecard --days 14".

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Invoker Bot
invoker: wf-1791728212537-13/publish-scorecard — Review claim: report…
923250d 
….py --html folds runs.jsonl and events jsonl once into summary.json and emits a static dashboard.html on demand, with no server and no change to any hook behavior.

Review lane: behavior
Safety invariant: Hook stdout and exit codes stay the same. This slice does not change the other four claims.
Files: engine/hooks/_runner/report.py
Change types: bugfix
Acceptance criteria: The branch contains only this claim, and the review step publishes one pull request.

Exit code: 0
Invoker-Finalize-Id: e5a0edf3-e934-4a17-8242-75988f61e013
@EdbertChan
Merge experiment/wf-1791728212537-13/publish-scorecard/g0.t0.a-aad9ca…
f74048a 
…478-adac8c27 — Review claim: report.py --html folds runs.jsonl and events jsonl once into summary.json and emits a static dashboard.html on demand, with no server and no change to any hook behavior.

Review lane: behavior
Safety invariant: Hook stdout and exit codes stay the same. This slice does not change the other four claims.
Files: engine/hooks/_runner/report.py
Change types: bugfix
Acceptance criteria: The branch contains only this claim, and the review step publishes one pull request.
Invoker Bot CI Bot
Run the SubagentStop hook fleet through one dispatcher process.
005c476 
The fleet self-saturates: one SubagentStop burst spawned 15-30 processes
that each re-parsed the same transcript, measured at 15.1s p50 wall idle
against 1.55s for a single process, with per-hook durations escalating in
launch order.

dispatch.py now runs opted-in detectors in-process, sequentially, parsing
the event once and passing the same parsed payload to every detector. A
hook opts in with subagent_stop.dispatch in its manifest, naming a
MODULE:FUNCTION entry point (default detect:detect). A detector with no
resolvable entry point still runs as a subprocess and its metrics row is
marked subprocess_fallback. Each detector keeps its own verdict, its own
runs.jsonl row under its own hook name, and its own rule_ids; a detector
that raises, exits, or overruns its 4s soft budget is caught and recorded
without stopping its siblings. The event budget defaults to 9s for
SubagentStop. mirror_stop_hooks_to_subagent_stop.py emits one dispatcher
entry for the opted-in hooks and leaves every other hook's direct entry
alone.

(cherry picked from commit 531f5c8 on
plan/hook-fleet-dispatcher-measure-the-herd-then-dispatch-it)
Invoker Bot
invoker: wf-1791728212829-14/publish-pilot — Review claim: SubagentSt…
6a3c5a1 
…op hooks dispatch through one process that parses the event once and shares one transcript parse across all opted-in detectors, emitting the same per-detector metrics rows and verdicts as the subprocess fleet.

Review lane: behavior
Safety invariant: Hook stdout and exit codes stay the same. This slice does not change the other four claims.
Files: engine/hooks/_runner/dispatch.py
Change types: bugfix
Acceptance criteria: The branch contains only this claim, and the review step publishes one pull request.

Exit code: 0
Invoker-Finalize-Id: aaa80940-f2d5-4d79-bb40-79c68fe333ad
@EdbertChan
Merge experiment/wf-1791728212829-14/publish-pilot/g0.t0.a-a268cef98-…
67ed846 
…e5bede34 — Review claim: SubagentStop hooks dispatch through one process that parses the event once and shares one transcript parse across all opted-in detectors, emitting the same per-detector metrics rows and verdicts as the subprocess fleet.

Review lane: behavior
Safety invariant: Hook stdout and exit codes stay the same. This slice does not change the other four claims.
Files: engine/hooks/_runner/dispatch.py
Change types: bugfix
Acceptance criteria: The branch contains only this claim, and the review step publishes one pull request.
@EdbertChan
Merge pull request #1330 from EdbertChan/stack/EdbertChan/reflect/les…
cfbdf1a 
…son-route-already-covered/drop-lesson-router-comments-failed-comment-gate--0a048d4d

Drop the lesson-router comments that failed the comment gate.
@mergify @mergify-commit-signer
Merge of #1324
9d24f21
@mergify @mergify-commit-signer
Merge of #1331
5337ffa
@mergify @mergify-commit-signer
Merge of #1334
31a3a61
@mergify @mergify-commit-signer
Merge of #1335
d195968
This was referenced Oct 11, 2026
Merged
Merged
Merged
Merged
@mergify mergify Bot closed this Oct 11, 2026
@mergify
mergify Bot deleted the mergify/merge-queue/924c40fec7 branch October 11, 2026 20:23
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@EdbertChan