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
Make human-merge-path holds attributable so the pin-only exception can clear only its own stale hold once a later exact head qualifies, and finish operator docs for the Utah rollout (projectbluefin/utah#1082).
Changes
HoldForHumanMergePaths already stamps <!-- hive-human-merge-path -->. Add the head SHA to that marker (keeping the old marker string matchable) so the hold's origin and head are known.
When a PR in a pin-only-enabled repo later qualifies on its current exact head (relay and sweep), remove holdonly if the sole hold source is a Hive human-merge-path marker comment and no human added/re-added hold after it (check label timeline events: actor must be the App). Never clear a human, security, or unrelated hold. Unknown timeline → keep hold.
Docs: src/docs/hive-merge.md + src/docs/operator-reference/README.md — full auto_merge.human_merge_pin_only reference, config reload path, minimum Hive version, and that the exception never bypasses required checks, freshness, reviews, or soak.
Tests
App-only path hold + qualifying pin head → hold removed; human-applied hold → kept; hold re-added by human after marker → kept; non-qualifying head → kept; timeline fetch error → kept.
Changelog fragment changelog.d/added-pin-only-hold-attribution.md.
🔒 Claimed by @mendezr (contributor on hive hosted-available-oke-11-placeholder-r05x). The claim lasts 30m0s without activity and renews while the work continues. Reply /unclaim to release, or /claim to take it over if you outrank the holder.
Opened partial implementation PR #16293 against v5: exact-head attribution, a shared fail-closed release API, regression tests, and operator rollout documentation. Existing/human/re-added/ambiguous holds are retained. Automatic relay/sweep release is intentionally not enabled: upstream/v5 currently lacks the schema/classifier and admission APIs from #16262, #16265, and #16266. After those land, admission must call ReleaseHumanMergePathHold only on a qualified exact head and re-read labels. This issue remains open for that integration and the concrete released-version/live Utah rollout evidence.
@hivecommons-hive[bot] this issue was closed by @clubanderson 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 — 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 — 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 2, 2026
🔒 Claimed by @mendezr (contributor on hive hosted-available-oke-11-placeholder-r05x). The claim lasts 30m0s without activity and renews while the work continues. Reply
/unclaimto release, or/claimto take it over if you outrank the holder.mendezr commentedon Nov 2, 2026
Opened partial implementation PR #16293 against v5: exact-head attribution, a shared fail-closed release API, regression tests, and operator rollout documentation. Existing/human/re-added/ambiguous holds are retained. Automatic relay/sweep release is intentionally not enabled: upstream/v5 currently lacks the schema/classifier and admission APIs from #16262, #16265, and #16266. After those land, admission must call ReleaseHumanMergePathHold only on a qualified exact head and re-read labels. This issue remains open for that integration and the concrete released-version/live Utah rollout evidence.
— hive: backend=pi model=openai-codex/gpt-6.1-sol
🐝 Hive Agent:
contributor| SHA:unknownhivecommons-hive commentedon Nov 2, 2026
🔓 Claim by @mendezr released (lease revoked).
🐛 fix(github): attribute human-merge path holds and document pin-only…
github-actions commentedon Nov 2, 2026
@hivecommons-hive[bot] this issue was closed by @clubanderson 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 — 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 — 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.