multica

mirror of https://github.com/multica-ai/multica.git synced 2026-07-14 05:39:08 +02:00

Files

J fcd6a0f4c1 MUL-2766: fix(squad): wake parent on child in_review, not just done

The runtime brief tells child agents to move their issue to `in_review`
when finished (the leader/human promotes to `done` after review). The
child-done notification was hardcoded to fire only on the `done`
transition, so squads were getting stuck: leader assigned work out,
child agents completed and moved to in_review, parent leader was never
woken.

Widen the trigger to fire on the transition into either `in_review` or
`done` (autopilot already treats them as a single completion signal).
Reuse the existing idempotency rule: once the parent has been told the
child is complete, the in_review ↔ done step is a no-op; a real reopen
(in_progress) followed by a fresh completion fires again.

Use different wording per status — in_review asks the leader to inspect
and accept/send back, done carries the existing promote-backlog-sibling
guidance.

Co-authored-by: multica-agent <github@multica.ai>

2026-05-28 13:09:46 +08:00

cmd

MUL-2744: feat(auth): auto-renew daemon PAT in-place within 7-day window (#3360 )

2026-05-27 22:22:26 +08:00

internal

MUL-2766: fix(squad): wake parent on child in_review, not just done

2026-05-28 13:09:46 +08:00

migrations

MUL-2703: feat(autopilots): webhook event filters per trigger (MUL-2334 follow-up) (#3231 )

2026-05-27 15:47:36 +08:00

pkg

MUL-2708: fix(agent): preserve multi-line Pi prompt on Windows by bypassing the .cmd shim (#3417 )