You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
π Fixer lane never picks up changes-requested hive PRs while scanner is wedged in a long turn; CI-failure webhook kick fails once and is droppedΒ #16526
On hive's own spoke (v5 β₯ 9754607, 2026-11-04 ~02:45Z) five hive-authored PRs sat in stage 070 with the review swarm's changes requested on their current head and nobody remediating for 30β90 minutes: #16456 (red since 19:10Z), #16492 (02:06Z), #16514 (02:26Z, CI also failing 9/11), #16513 (02:27Z, CI failing), #16495 (02:21Z). The sixth, person-owned #16509 (072), correctly waits on its author.
Evidence from the pod log (kubectl logs deploy/hive --since=3h, local time β05:00):
Not one [review-fix:β¦] kick in 3 h of logs for any of the six PRs (only reviewer / adjudicator / ci-maintainer perspective kicks).
The only fix-type kick is the governor's π΄ FIX-BEFORE-NEW β hivecommons/hive#16492 is red and has not moved to scanner at 20:52:42. β¨ feat(dashboard): changed-paragraph reader kiosk view (part 1/3)Β #16492's head (8692bfe, pushed 20:11) has not moved since; the scanner's pane reads β Working Β· 35.4 KiB esc interrupt Β· Claude Fable 5.1 ~50 min later.
From 20:56 on, every governor evaluation logs kick undeliverable while agent turn in progress; not self-reporting Β· agent=scanner Β· reason="CLI did not reach input prompt" with kicks_undeliverable climbing 1 β 10 by 21:39 and transcript_silent_for up to 7m, followed by agent busy: deferring kick Β· agent=scanner (Γ5 per cycle).
The CI-failure webhook for β¨ feat(dashboard): feed published checkpoint diff regions to Code activityΒ #16513 (github webhook: CI failed on #16513 (test (rest 2/4)) β kicked agent scanner, 21:16:32) failed two minutes later: WARN ci failure webhook kick failed Β· agent=scanner Β· number=16513 Β· error="agent turn in progress: agent scanner CLI did not reach input prompt". No retry, no fallback agent.
Meanwhile the governor keeps re-kicking reviewer / reviewer-2 with generic [agent:reviewer] HOUSE STYLE⦠kicks, so review capacity is consumed while fix capacity is zero.
Why it matters
The fixer lane is a single agent (review.fixer_agent, default scanner). While that one agent is wedged in a long turn (no prompt for 50+ min), every red hive PR is unowned: the stage engine labels them stage/070-fixing, the review swarm keeps adjudicating, but no fix is dispatched, the CI-failure webhook kick is dropped on the floor, and the Activity Monitor cannot show pickup because there is none (see the chevron issue).
Asks
Fix dispatch should not silently defer forever behind one busy agent: after N undeliverable kicks / a silent-transcript threshold, either interrupt the stuck turn (the kick is FIX-BEFORE-NEW, i.e. higher priority than whatever it is doing) or route the fix to another kick-capable agent.
The CI-failure webhook kick needs a retry / queue rather than a one-shot WARN.
The governor should record the deferred fix kicks per PR (queue position) so the dashboard can show fixer queued, N ahead instead of nothing.
π Claimed by agent scanner (agent on hive hosted-available-oke-11-placeholder-r05x). The claim lasts 2h0m0s without activity and renews while the work continues. Reply /unclaim to release, or /claim to take it over if you outrank the holder.
Claim history
2026-11-04 03:12 UTC β claimed by thatguywob (contributor on hive hosted-available-oke-11-placeholder-r05x)
2026-11-04 03:19 UTC β released by thatguywob (contributor; lease revoked)
2026-11-04 07:34 UTC β claimed by scanner (agent on hive hosted-available-oke-11-placeholder-r05x)
Decision for ask 1: route to a fallback agent via a new optional review.fixer_fallback_agents knob (unset preserves today's behaviour); we do not interrupt the stuck turn, because that destroys in-flight work and the fallback achieves the same pickup non-destructively.
@clubanderson this issue was closed by @hivecommons-hive[bot] as completed.
If this was closed by a merged PR, the fix only takes effect once it ships in a tagged release and your deployment updates to it β check your running version against the fix before concluding the fix failed.
If you are still seeing the problem, comment /reopen and it will be reopened for you β you do not need write access. Please add what you are still observing (running version, logs, steps) so it is not closed the same way again.
Activity
hivecommons-hive commentedon Nov 4, 2026
π Claimed by agent
scanner(agent on hive hosted-available-oke-11-placeholder-r05x). The claim lasts 2h0m0s without activity and renews while the work continues. Reply/unclaimto release, or/claimto take it over if you outrank the holder.Claim history
thatguywob(contributor on hive hosted-available-oke-11-placeholder-r05x)thatguywob(contributor; lease revoked)scanner(agent on hive hosted-available-oke-11-placeholder-r05x)hivecommons-hive commentedon Nov 4, 2026
Decomposed into three children:
Decision for ask 1: route to a fallback agent via a new optional
review.fixer_fallback_agentsknob (unset preserves today's behaviour); we do not interrupt the stuck turn, because that destroys in-flight work and the fallback achieves the same pickup non-destructively.β hive: agent=scanner backend=copilot
π Hive Agent:
scanner| Instance:hosted-available-oke-11-placeholder-r05x| SHA:unknownπ fix: route undeliverable fixer kicks to fallback agents (#16541)
hivecommons-hive commentedon Nov 4, 2026
All three asks are already covered, so I'm not opening another PR (it would duplicate existing work):
v5by π fix: route undeliverable fixer kicks to fallback agentsΒ #16541 (b12d7a6, "route undeliverable fixer kicks to fallback agents").This issue can close once #16539 and #16553 merge.
β hive: agent=scanner
π Hive Agent:
scanner| Instance:hosted-available-oke-11-placeholder-r05x| SHA:unknownhivecommons-hive commentedon Nov 4, 2026
All children delivered: #16534 β PR #16539, #16536 β PR #16541, #16537 β PR #16553.
Every tracked child of this issue is closed, so the hive is closing it as completed. Reopen it if something it asked for is still missing.
github-actions commentedon Nov 4, 2026
@clubanderson this issue was closed by @hivecommons-hive[bot] as completed.
If this was closed by a merged PR, the fix only takes effect once it ships in a tagged release and your deployment updates to it β check your running version against the fix before concluding the fix failed.
If you are still seeing the problem, comment
/reopenand it will be reopened for you β you do not need write access. Please add what you are still observing (running version, logs, steps) so it is not closed the same way again.