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
Seen on hive's own hosted spoke, 2026-10-28 16:20Z, Activity Monitor with the Needs human 18 chip active.
Bug 1 — "Needs human" filter includes rows no human needs to act on
Visible under the filter: #14976 at 040 "PR in progress" (modifiers [], reason "open PR covers it"), #7563/#14939 "Ready for auto merge" + hold pill, #2364/#9699/#9701/#9735 "Open · hold".
Cause: the chip is label-based — stageTrackerItemNeedsHuman(item, set) matches item labels against the payload's needs_human_labels (the Projects preset set from needsHumanPresetLabels: hard-suppress buckets, needs-signal, hold labels, 071, unauthorized, sentinel, needs-triage). #14976 wears needs-signal; the others wear hold. There was no server-side needs_human count, so the chip and filter could not be checked against the API either.
Fix: "Needs human" = rows where a human decision/action is the only thing that advances the item — stages 015, 045, 085, 090 (ready-for-human-merge), 071/072 when the author is a human, and the needs-human / needs-decision / unauthorized modifiers. Not hold (→ new ⏸ On hold chip), stalled, blocked, handled regressions, 040 covered, ready-for-automerge. One predicate (stageTrackerNeedsHuman Go ↔ stageTrackerItemNeedsHuman JS) feeds the chip count, the filter, and new payload fields needs_human / on_hold (+ per-item flags). One-line tooltips on both chips.
Bug 2 — #14976 (the [shared-ci] incident issue) shown at 040 "open PR covers it"
Thesis was that the "Deferred to shared-CI incident #14976" comments on #14939/#14974 created a covering relation. The live tracker history says otherwise: #14976 went 030 → 040 at 12:16:38 ET, linked PR #14990 (hive's scanner fix, body Refs #14976 (needs-human: …), no closing keyword). The relay (pr_request_watcher → clearInProgressForPR(prCoveredIssues(...))) covered the declared issue regardless of the explicit non-closing Refs, and the sweep's Reference claim path then stated the same reason. Timeline cross-reference events are never used to derive coverage (verified), and ParseClaimedIssues/ParseReferencedIssues ignore "Deferred to shared-CI incident #N" (regression test added).
Fix: a reference-only (Refs) claim is carried as ClaimFact.Partial; the relay/enrich/backfill paths record it via a new FactSink.AddressedBy, and the engine/dashboard say "open PR #N addresses part of it (references it without a closing keyword)" instead of "covers it". Coverage requires a closing reference, a ledger closing claim, or an explicit marker.
Follow-up (not in this PR): [shared-ci] incident issues are detected only by title prefix (sharedCIIncidentTitlePrefix); there is no label/marker to give them a dedicated "incident" presentation in the tracker.
@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 — 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
dashboard: port label_facets.test.js to the stage/modifier Needs huma…
github-actions commentedon Oct 28, 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 — 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.🐛 dashboard: Needs human chip counts only human-gated rows; Refs-only…