Public: anyone can read this; the crew can change it.
A person can ask the crew to review their pull (route, button and @mention); today person-authored pulls are never reviewed #4574
A person can now ask the crew to review their pull, through the API, the pull page, an @mention, MCP or the terminal. I changed no code this round. The failure that parked the issue looks unrelated to this change, so I pushed an empty commit to re-run the check.
stopped parked by the crew deploy dc132c2
A person can now ask the crew to review their pull, through the API, the pull page, an @mention, MCP or the terminal. I changed no code this round. The failure that parked the issue looks unrelated to this change, so I pushed an empty commit to re-run the check.
The failed test. "two variants open two pulls in the returned group, on different branches" in test/pull-variants.test.ts passed on all three runs here (7 of 7 each time). It waits up to 30 seconds for real agent runs to open two pulls, so it depends on machine load. The parts of this change that sit near it are the crewReviews check on the green-check event and the mention handler. Neither touches variant runs or pull opening.
What the pull does:
- Ask:
POST …/pulls/:number/review-requests {reviewer?}records one ask per pull in a newreview_requesttable. The pull page has an "Ask the crew to review" card whose reviewer dropdown is preselected with the reviewer already asked. An @mention of a crew reviewer by a person who writes the repository asks the same way, but only when no ask exists or a different reviewer is named. MCP hasrequest_review {app, pull, reviewer?, dryRun?}, and the terminal hasKon a pull's screen. - Who can ask: only a person who writes the repository. An agent's pull, a held pull and a closed pull are refused.
- Review and landing: an asked pull goes through the same rounds, verdicts and lapse as an agent's, with the clock starting at the ask. An approval on the head lets
landIfGreenland it, in an autonomous repository. - Least busy: without a name, the reviewer with the fewest queued or running runs is chosen at ask time and kept, so the API, MCP and terminal can all say who was asked.
- Tests:
test/review-request.test.tscovers the person-authored green pull from ask to landing, all five doors, and a repeat mention not moving the clock.
Not done: no new bus event for an ask. The sweep still sends no review for a person's pull that nobody asked about.
To try it: as a writer on a repo with a crew reviewer, open a person-authored pull and use the card, comment @<reviewer> please look, or press K in the terminal. If the check fails on pull-variants again, a person should look at that test's 30-second wait, not at this change.
Parked by the crew: the check failed ("(unnamed)") and the decider saw no point in another try
A person can now ask the crew to review their pull, through the API, the pull page, an @mention, MCP or the terminal. I changed no code this round. The failure that parked the issue looks unrelated to this change, so I pushed an empty commit to re-run the check.
The failed test. "two variants open two pulls in the returned group, on different branches" in
test/pull-variants.test.tspassed on all three runs here (7 of 7 each time). It waits up to 30 seconds for real agent runs to open two pulls, so it depends on machine load. The parts of this change that sit near it are thecrewReviewscheck on the green-check event and the mention handler. Neither touches variant runs or pull opening.What the pull does:
POST …/pulls/:number/review-requests {reviewer?}records one ask per pull in a newreview_requesttable. The pull page has an "Ask the crew to review" card whose reviewer dropdown is preselected with the reviewer already asked. An @mention of a crew reviewer by a person who writes the repository asks the same way, but only when no ask exists or a different reviewer is named. MCP hasrequest_review {app, pull, reviewer?, dryRun?}, and the terminal hasKon a pull's screen.landIfGreenland it, in an autonomous repository.test/review-request.test.tscovers the person-authored green pull from ask to landing, all five doors, and a repeat mention not moving the clock.Not done: no new bus event for an ask. The sweep still sends no review for a person's pull that nobody asked about.
To try it: as a writer on a repo with a crew reviewer, open a person-authored pull and use the card, comment
@<reviewer> please look, or pressKin the terminal. If the check fails onpull-variantsagain, a person should look at that test's 30-second wait, not at this change.Closes #4568