Repository navigation
Description
Why
Owner (2026-10-28 12:11 ET, verbatim):
Hive is doing the right things mostly; we now need to keep the user informed of it — like a video game they can watch and see what happens easily.
Right now I look at these rows and want to take action, but I need to know what my options are; once I act I need to see that I acted and what I chose; I need to see that work is in progress. Throughout the Activity Monitor we need visual cues for (a) current state, (b) whether work is happening in that state, (c) what decisions humans and hive made. This has to be exceedingly clear, easy to read, and concise.
Owner (12:14 ET): the game is a mental model for what must always be legible — position, moving vs waiting, what's blocking, transitions, failures, completion — not a visual theme. Professional, enterprise-acceptable; plain legend words (working / waiting on X / set back by Y / merged); 150–300 ms transitions, one gentle highlight on change, understated merge celebration.
Method
Live, read-only walkthrough of hive's own hosted dashboard (hive.hivecommons.dev, Activity Monitor, 40–47 rows) through the operator's Chrome tab via CDP, four passes: 12:11 and 12:29 ET on 0470256, 14:17 on 9627b20, 16:59 on aa9b530 (after #14995 #15013 #15014 #15011 #15016 #15056 #15060 #15063 #15080 #15084 rolled). Each pass: full-board screenshot, every chevron / pill / age / tooltip / ⋯ menu read as a human would, graded per row on
(a) can I tell the state and why · (b) is hive working on it right now vs waiting, on what · (c) what decisions were made, by whom, when · (d) what can I do next and what will it cause · (e) 2-minute watch test: if I just watched the screen, would I know what hive did?
Spot-checked against GitHub (gh pr view … --json labels,state,mergeStateStatus, ≤1 call per row per pass). Full per-row tables and screenshots for every pass are in the operator's ~/hive-work/op-eyes/<ts>/findings.md.
Consolidated findings — status after aa9b530
Legend: ✅ fixed on the live board · 🟡 partially / shipped-but-not-observed · ❌ open · PR = covered by a merged/open PR (checked by number).
| # | Finding (exact UI text) | Grade | After roll | Covered by | Suggested |
|---|---|---|---|---|---|
| G1 | No cue that hive is working on a row. 16:59: top bar scanner RUNNING NOW, adjudicator RUNNING NOW ×2, reviewer queued UP NEXT, yet 0 of 47 rows carried cue-working/cue-acting. Fixing · 5h (#14921), Fixing · 12h (docs#500), Soaking, Ready for auto merge all render a static chevron. |
b ❌ e ❌ | 🟡 | #15014 shipped the spinner/stripes, but they fire only on operator actions | Drive the working cue from hive's own activity (SSE already consumed by the tracker): working: adjudicator · 40s, reviewer queued · 2nd. New scope. |
| G2 | Red "needs human" breathe (data-motion=needs-human) on rows nobody is waiting for a human on: Open · blocked · since 3h (#15030 #15045 #15047 #15089 — tooltip: "lifts when PR merges"), Actionable · blocked (#14636), Actionable · stalled (#14598), Returned to contributor · Next move: clanker (#15064). |
b ❌ | 🟡 chip count fixed (Needs human 13, was 19) |
#15013 fixed the chip | Breathe only when Next move: human; blocked/returned rows get a static amber "waiting on #14945 / @login". New scope. |
| G3 | Age has no tooltip and no stable meaning. since resets on every hop (#14976 bounced Actionable → CI/Review → Actionable → Needs human in 2h, each time now), is blank on all Open · hold rows, and disagrees between issue and PR rows (#15009 3h vs 1h; #15051 2h vs 25m). |
a |
❌ | — | Tooltip "in this stage since 01:58 · last hive action 11:44 (nudge #2)"; always render age; +N hops marker when a row has bounced. |
| G4 | Pills without tooltips: hold +1, needs rebase, regressed +1, stage ↓, v5/v6/main, tracker 2/4, needs human (mod pill). hold never says who/when/why. +N never expands. |
c ❌ d |
🟡 (waiting on @x (online, 1 nudge) and stalled +2 now have tooltips) |
#15070/#15054 (presence on waiting-on pill) | hold · @clubanderson · Oct 12 · "wait for v6 epic"; expand +N on hover. |
| G5 | Decision trail has no actor / time / reason — only Trail: Actionable → Claimed → Actionable → Covered by PR inside the issue-number tooltip. No row at 16:59 showed a decision badge although Merged today lists actions minutes earlier. |
c ❌ e ❌ | 🟡 shipped, not observed | #14992 (badge shows what you chose), #15003 (closed) | Row-level trail chips Claimed by scanner 09:55 → Returned (conflicting) 10:42 → nudge ×2 → adopted 12:46. |
| G6 | ⋯ menu items state no consequence: Hold, Re-evaluate now rescan, Mark needs-human…, Let hive decide delegate, Approve (/lgtm + /approve), Give direction…. Decide PR #14822… ▸ submenu does not open on hover. Hold offered on a row already on hold (#7563). |
d ❌ | ❌ | #15085 / #15087 add actions, not explanations | One-line consequence under each item: "adds hold to #14822; hive stops touching it until released". Hide Mark needs-human… / Hold when already in that state. |
| G7 | No post-action acknowledgement visible on the row (what I chose, when). | c ❌ e ❌ | 🟡 shipped | #15014 toasts, #14992 badge | Verify on next operator action; keep the last event per row (see N3). |
| G8 | Stage headers (Merge gate), the ✓ column and since header have no tooltips. |
a |
❌ | — | Header tooltips; label the ✓ column "Merged". |
| G9 | Meta updated 2s ago · stale (contradictory; was · fresh at 14:17, back to · stale at 16:59); anomalies chip with no count. |
a |
❌ | — | "cache 2s · GitHub data 4m old"; "0 anomalies". |
| G10 | Jargon in the only "why" text: review gate: no review bundle, adoption stalled: contributor lane: workflow path, 085 park is overwritten by 071/072, 085 tooltip lists three possible causes ("sentinel alert, hold/needs-decision label or human-only rule") instead of the actual one. |
a |
🟡 (090 text and the hold popover are plain English ✅) | #14995 (popover), #15063 (plain-words decision outcome) | Name the actual cause and actor: "needs-decision set by reviewer @x at 10:31: approve-with-rationale". |
| G11 | Rows that finish vanish without an end state. Merges now celebrate ✅, but PRs closed unmerged / superseded (PR #14979, PR #14939 off #7563) simply disappeared between passes. | e ❌ | 🟡 | #15014 / #15060 (merge only), #15011 (auto-close superseded) | Brief Closed · superseded by #N / Closed unmerged end state before the row leaves the board. |
| G12 | stage ↓ (#15061), regressed +1 (#15064) markers have no from→to / reason; cue-setback class present on 3 rows but renders nothing visible. |
a |
🟡 (stalled +2 now shows "080 Merge gate → 070 Fixing at 04:26 because conflicting" ✅) |
#15014 (setback event line) | Same from→to+reason text on every setback marker, visible on the row not only in a tooltip. |
New after aa9b530
| # | Finding | Grade | Covered by | Suggested |
|---|---|---|---|---|
| N1 | Needs-human popover does not open on stage-085 rows. Clicking Needs human decision 👤 (button.ham-decision-trigger[aria-haspopup=dialog]) on #14816, #14633, #14976: aria-expanded stays false, no .ham-decision-pop, no JS error. It works on Open · hold rows (#13216: "Should the hive work on this issue? … Accept triage / Release hold / Mark needs-human"). |
d ❌ | regression in #14995 — not covered | Fix the trigger→popover wiring for 045/085/090 (likely issue-key vs PR-key on nested rows). |
| N2 | The chevron lost its hover text: title="", the text moved to data-ham-tip which nothing renders. Hovering Needs human decision now shows nothing (regression vs. 12:11). |
a |
not covered | Render data-ham-tip through the custom tooltip or restore title. |
| N3 | cue-event line ("Hive: unblocked · 1m ago") appeared once and faded; 10 min later zero event lines on 47 rows despite 5 merges in 40 min. A spectator who arrives late sees nothing. |
e |
#15014 (fades by design) | Keep the last event per row persistent (dimmed, with age); fade only the highlight. One line per row. |
| N4 | Ready for auto merge (#15083, #15088) shows no queue position / soak timer / gate it waits on — visually identical to Soaking. Soak countdown ("23m left") exists only in the tooltip. |
b |
#15084 (vocabulary) partially | Queued · behind 2 · soak 23m left on the chevron. |
| N5 | Fixing with no fixer evidence for 5h / 12h; the stalled pill was removed from #14910/#14911 while nothing changed. |
b ❌ | not covered | After N min without a fixer run: Fix not started · 5h + Re-kick action. |
| N6 | Top-bar agents (adjudicator RUNNING NOW) are not linked to rows — no row can say "adjudicator is on this". |
b ❌ e ❌ | not covered (same root as G1) | Per-row working cue from agent kicks. |
Contradictions with GitHub seen during the walkthrough
- 12:11–14:17: 📝 v6 epic + readiness bar: dashboard-optional operation — meet humans where they are #7563 → PR 📖 docs(v6): behind fork leaves the lane front at once; serialized lane readiness row from the dd406b3 re-run #14939 showed
Ready for auto mergefor >2h while GitHub hadstage/060-reviewed, mergeStateUNSTABLE, and theholdlabel was on the issue. Its own tooltip said "not yet admitted — a gate is still pending — deferred to shared-CI incident [shared-ci] Go module proxy stream resets (INTERNAL_ERROR) failing golangci-lint, overlayfs-exec-guard and docker (hub) across hivecommons/hive #14976". (🐛 Activity monitor: one merge-gate vocabulary and a labelled leading glyph on every gated row #15084 unified the vocabulary; worth a regression test.) - 12:11: PR 🐛 fix: preserve supersession decisions across stage reconciliation #14979 showed
Needs human decision(085) while GitHub hadstage/050-review-pending,BLOCKED. - All other sampled rows (🐛 fix: preserve owner intent through PR decision paths #14822, 🐛 fix: preserve supersession decisions across stage reconciliation #14821, ✨ feat(hub,agent): Hub-coordinated live CLI session pool for multi-tenant runner capacity #14524, 🐛 fix: release label-blocked serialized lane fronts during sweeps #14945, ✨ review: gate PR reviews on an approval label (review.require_labels) #14987, ✨ feat: let a sandboxed reviewer post reviews through a workspace spool #14989, ✨ feature: inline the linked Linear issue into review kicks #14921, docs#500) matched their labels.
Proposed acceptance test (the 2-minute watch)
Open the Activity Monitor and don't touch it for two minutes. For every row that changed you should be able to say, without hovering or clicking: where it is, whether hive is working on it (and which agent), what it is waiting on, what just happened to it (who/what, how long ago), and — if it left the board — how it ended.




Activity
clubanderson commentedon Oct 28, 2026
Iterations 5 (17:14, aa9b530) and 6 (17:17, after roll 2 → dd412e4)
50 rows. Meta
updated 1s ago · stale. Top bar: reviewer queued UP NEXT, reviewer, scanner. ChipsNeeds human 12,On hold 12,Merged today: 33(#14701 17:06, #14989 16:56, #15042 16:48).Screenshots: 00-activity-monitor-full.jpg, 01-viewport.jpg.
2-minute-watch (16:59 → 17:14)
@hivecommons-hive[bot], author/hive) labelled "untrusted author"; vocabulary flipped within 15 min with no event line. GH label 090 matches UI, so the stage derivation is wrong, not the display. New N7.needs rebasepill vanished; age unchanged although state changed (G3).stage ↓stage ↓vanished silently; soak countdown only in tooltip (N4).cue-setbacktext, no reason on row (G12).cue-eventfading: "Hive: Hive adopted this PR; hive is addressing the review findings — review swarm requ…" 1m ago; gone 60s laterreviewer queued UP NEXTGlobal re-grade (unchanged from iteration 4 unless noted)
G1 ❌ · G2 🟡 · G3 ❌ (confirmed again: #15051) · G4 🟡 · G5 ❌ (#14633, #15036 decisions invisible) · G6 ❌ · G7 🟡 · G8 ❌ · G9 ❌ (⚠️ (event line verbose/duplicated prefix).
stale) · G10 🟡 · G11 ❌ (#14701 merged → row regressed instead of ending) · G12 🟡 (#15025 setback silent) · N1 not re-tested · N3New
Operator-eyes — iteration 6 — 17:17 ET — AFTER ROLL 2 (dd412e4) (#15056/15060/15063/15080/15084 live)
Banner: "Upgraded aa9b530 → dd412e4: 9 unreleased changes." 50 rows; chips Needs human 11 / On hold 12 / Merged today 33. Tab picked up the new frontend without reload.
Screenshots: 00-activity-monitor-full.jpg, 01-viewport.jpg.
Re-grade after roll 2
cue-workingrows = 0, spinners 0, while top bar showsscanner reviewer revieweractive.sinceempty; #15095Open · since 18mbut age column blank.needs human decisionchip, no actor).stalemetastage ↓/stalled markers w/o tooltip.ham-decision-pop. Control: #13216 (hold) and #14976 (045) open fine. Operator must use ⋯ → "Decide PR …▸" instead; nothing tells them that.title="", tip only indata-ham-tipGood after roll 2
Still the two biggest gaps for the "watch for 2 minutes" test
hivecommons-hive commentedon Oct 28, 2026
🔒 Claimed by @Karibhou (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.hivecommons-hive commentedon Oct 28, 2026
🔓 Claim by @Karibhou released (lease revoked).
clubanderson commentedon Oct 28, 2026
Iteration 7 (17:48 ET, dd412e4) — did the merge celebration fire for #14945 (17:18) / #15067 (17:42)?
Facts read from the live tab (read-only, CDP)
v5.186.4-839-gdd412e4, banner "Upgraded aa9b530 → dd412e4");navigation.type = reload, document loaded 17:16:07 — i.e. the tab self-reloaded after roll 2 and was on the new build before both merges. Not a stale pre-roll tab.hive.tracker.celebrateunset → default on (⚙ → Merges checkbox checked).prefers-reduced-motion= no. Sound:hive.tracker.sound=0(off) → no zap either way._stageTrackerZapped= {pr#14945, pr#15067} → both merges were detected by the SPA._stageTrackerPRHostmaps pr#14945 → issue#14934, pr#15067 → issue#15061 (host rows resolved, socelebrateRowhad a row for each).merged_24h/audit seed lost.)cue-mergedrows 0,cue-token0,_stageTrackerCelebrating{} — expected, the effect is 2.5 s pin + 0.52 s flight and the event line lives ≤2 min. No persistent trace exists after ~2 min by design.#activity-monitor-mergedsits in.ham-toolbar(position: static). Operator's tab:scrollY = 10423, chip attop = −2295 px, first row at−2139 px→ the counter, the glow target and any token landing are off-screen when scrolled into the list, which is where an operator reading rows is.⚙ → Merges → Test celebration (slowed ×6 via an injected stylesheet so CDP frames could catch it; removed after)
top: −2111px. The operator sees nothing when pressing the test button (10-settings-merges.jpg, 11-test-frame*.jpg). Closing the dialog and re-running: still off-screen (12-test-nodialog-frame*.jpg). New N9.cue-mergedapplied to 🐛 decision path: approve-with-rationale reads as changes_requested, Refs→Fixes rewrite overrides owner decision, 070 deadlock without a new head #14816, count unchanged as documented).Verdict
The celebration would NOT have been reliably visible to an operator with the tab open since 17:13, because: (1) it is a one-shot 2.5 s glow + 0.5 s token with sound off and no persistent trace; (2) the target chip and the row are in a non-sticky toolbar/list — at the operator's scroll depth both the source rows and the "Merged today" chip were off-screen; (3) the Test button plays behind its own modal and at the top of the list, so even the self-check shows nothing. The detector and counter did fire (both PRs in the zapped set, count 2 → 4), so this is a visibility failure, not a detection failure.
Suggested fixes (for the celebration PR owner)
.ham-toolbar(or just the Merged-today chip)position: sticky; top: 0inside the section, so the token always has an on-screen landing and the count pop is seen.#15101 (merge-gate pill two-line clip) and merge-train rows
.ham-segtaller than one line, none clipped. Can't reproduce right now; iteration-6 capture (🐛 [v6] contributor-lane auto-promotion runs hive-wide searches per PR (bots included, no per-author memo) and trips GitHub's search secondary limit #15088 → 🐛 fix: limit contributor promotion searches per sweep #15090 "Awaiting merge gate · open PR 🐛 fix: limit contributor promotion searches per sweep #15090 covers it (PR at 060-reviewed)") stands as evidence.clubanderson commentedon Oct 28, 2026
Iteration 8 (18:02–18:10 ET, dd412e4) — adoption pairs after "Let hive decide" / Approve — N10–N12
Rows watched: #15023 (#15029 → #15123), #15083 (#15086 → #15124), #15088 (#15090 → #15125), #14816/#14822 (Approve 17:59), #14938/#14947 (Re-run decision). Polls at 18:02:30, 18:05:06, 18:06:22, 18:07:19, 18:07:58, 18:09:01, 18:10:03. Screenshots 20-, 22-, 23-*.
Timeline as seen on the board
_stageTrackerZappedhas pr#15124 (detector fired) but nocelebratingentry or event trace survived 20 s later. Original #15086 now appears as its own row: "Adopted → #15124 · 8m", GH: OPEN, stage/074#14816 (→#14822): "Needs human decision" +
needs rebase; ✓ badge tooltip: "You chose Approve · 05:59 PM EDT · done — your approval was posted as a comment." GH: #14822 OPEN, DIRTY, stage/085 — nothing moved in 11 min and nothing on the row says why (rebase needed → hive won't act / contributor must).#14938 (→#14947): "Needs human decision"; ✓ tooltip: "You chose Let hive decide · 05:56 PM · done — the hive decided: adopt-supersede (rule: contributor offline)." GH: #14947 OPEN stage/085, no copy PR after 14 min — the chevron still says the decision is owed although it has been made.
Grades
[role=status]. The ✓ badge tooltip does lead with outcome ("Addressed — now at Adopted by hive. You chose Approve …") — good, but it is hover-only (badge is a 20 px ✓).cue-working(2 rows: #15103, #15099). None on the adoption pairs even while CI ran on the copies.New findings
🐛 Persist a merged ledger under /data for the Activity Monitor Merged…
clubanderson commentedon Oct 28, 2026
Iteration 9 (18:49 ET, dd412e4) — #15009 / PR #15020 four minutes after "Let hive decide" — N13
What the row shows (literal): strip chevron "Needs human decision 👤" (red/needs-human pulse) in the Merge-gate column · age 4m · badges
refsclosed(issue #15009 struck through) · ✓ (green, 20 px) ·needs humanpill (amber outline) · no event line · no working cue · no sub-row.Hover texts:
GitHub reality (18:50): PR OPEN, reviewDecision CHANGES_REQUESTED, labels
needs-human+stage/085-needs-decision. Comments: 18:23 hanthor/hive decision …; 18:45 hive posts the delegated/hive decision make CI green without widening scope…and "Decision recorded by @clubanderson". So the decision is recorded, but theneeds-humanlabel was never removed → stage snapped back to 085 within the same minute.Grades (from the row alone)
needs humanpill, popover "A person must decide … Asked by the hive · 4 min ago".N13
After "Let hive decide" the hive records the decision and moves the PR to Fixing, but the
needs-humanlabel it was parked on stays → the next sweep re-parks it at 085 and the row re-asks the human. Two fixes: (1) backend — a delegated/recorded decision must clear theneeds-humanlabel it answered (or the stage derivation must treat a recorded decision newer than the label as answered); (2) UI — when the ✓ trail's latest entry is a hive decision, the row must say so without hover: event line "Hive decided: fix then merge — fixing hanthor's requested changes", chevron "Fixing", noneeds humanpill, popover closed or reading "Decided 18:45 by hive (delegated by you) — nothing needed from you". The ✓ hover should quote the direction text, not the rule id.clubanderson commentedon Oct 28, 2026
Iteration 10 (18:54 ET, dd412e4) — #14816/#14822 after Approve and #14938/#14947 after Let hive decide — N14
Literal row text
#14816 → #14822 (thatguywob, GH: CONFLICTING,
needs-rebase+stage/085, CHANGES_REQUESTED): chevron "Needs human decision 👤" · age 16h · badgesv5· ✓ ·needs rebase· no event line · no working cue..ham-decision-pop).#14938 → #14947 (GH: MERGEABLE/UNSTABLE,
stage/085, no review; earlier copy #14977 CLOSED): chevron "Needs human decision 👤" · age 1h · badgesv6· ✓ · no event line.What GitHub already says that the row doesn't
On #14947 the hive itself posted at 17:56: "Hive decided close this PR and open a fresh hive-authored PR with the same commits, but it hasn't taken effect. What should have happened: … Why it hasn't: Required checks are failing on this branch; nothing merges until they are green. What you can do: Merge as maintainer · Re-run decision · Close as superseded · Send back for a new commit." — exactly the plain-words explanation an operator needs, and none of it reaches the Activity Monitor row (which still says "decision requested"). On #14822 the last bot comment is Prow "PR needs rebase" (16:04); the hive never says what "adoption stalled: contributor lane: workflow path" means or that the rebase is the blocker.
Grades
needs rebasesays someone must rebase; menu offers 5 more decisions. Nothing says "wait — blocked on rebase by @thatguywob" or "rebase it / let hive adopt".Good: the new ✓ "Still at … N min after you asked — hive says … Options: …" nudge text is the right instinct — it just lives behind a 20 px hover and still ends by offering more decisions instead of naming the blocker.
N14
When a human action or a delegated decision is recorded but the item does not move, the row must switch from asking to explaining: chevron stays where it is but reads "Stalled after your Approve · needs rebase" / "Stalled after hive decided adopt-supersede · checks failing", the event line carries the hive's own stall comment ("Why it hasn't: required checks are failing"), and the primary row action becomes the single recommended one (Re-run decision / Merge as maintainer / Ask @author to rebase) with the rest under ⋯. Internal reason codes ("adoption stalled: contributor lane: workflow path", "decision requested") need a human sentence; the hive's GitHub stall comment already has one — surface it.
clubanderson commentedon Oct 28, 2026
Iteration 12 (19:25 ET, after roll 3 → d93adf9) — every row older than 30 min: moving or stuck, and on whom — N15
Two points: 18:56 capture (iteration 11, dd412e4) vs 19:25 capture (d93adf9; banner "Upgraded dd412e4 → d93adf9"). "Moving?" = chevron/sub-row state changed between the two, or a fresh event line. 20 rows qualify.
needs rebase· ✓stalled +2needs rebasetracker 0/3·needs rebaseneeds rebasedraftneeds rebase +1↳ #15149 · under #15117needs humanblocked on #15104blocked on #15134trackerCount: 20 rows > 30 min · moving 3 (#15117, #15141 re-nested, #14934 reset-to-New) · stuck 17.
Top 3 stuck reasons
holdlabel on the front PR never releases the front, so the lane is blocked until the front timeout #14934 label pass). d93adf9's "next hive pass in Nm · no one is working on this right now" text fixes this for New rows; Actionable/Adopted/CI-Review rows still lack it.N15
Stuck-row legibility after d93adf9: author-blocked rows are now clear; the two remaining gaps are (a) rows where a decision was recorded but did not take effect still ask rather than explain (see N13/N14), and (b) hive-queue rows (Actionable / Adopted by hive / CI / Review) do not carry the new "no one is working on this right now · next hive pass in Nm · gates ahead" sentence that New rows got. Also a fresh contradiction: #14934 regressed to "New · since 9h" with its PR #14945 merged 2 h earlier (issue never closed) — the row should read "PR #14945 merged 17:18 · issue left open — close or reopen work?".
🐛 Persist a merged ledger under /data for the Activity Monitor Merged…
🐛 Persist a merged ledger under /data for the Activity Monitor Merged…
🐛 Persist a merged ledger under /data for the Activity Monitor Merged…
🐛 fix: open the needs-human popover on issue rows with a parked PR (#…
github-actions commentedon Oct 29, 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.🐛 Activity Monitor: persist the merged count in a server ledger, pin …
clubanderson commentedon Oct 29, 2026
N16 — open > 2 h, not on hold (20:38 ET, roll 3 · d93adf9 still live;
/api/version: behind 15 commits, target 73cdc6d, next update 2026-10-29 17:00Z)Board: 39 rows, Since filter = Any age, Per page 50 (nothing hidden). 11
Open · holdrows ignored. 6 rows are older than 2 h by their "since" column.(a) waiting on a named human — 2
draftneeds rebase +1+1hides "draft"; rebase isn't mentioned in the hover)(b) waiting on hive, no worker / ETA — 1
tracker(c) decision recorded but not applied — 2 (the two oldest non-draft rows)
needs rebase· ✓2workflowspermission) — ask @thatguywob to rebase, or Close as superseded" instead of re-listing the same 5 options(d) contradictory / unclear — 1
blocked on #15104Cross-cutting
18hwhile its ✓ says the operator acted 37 min ago and hive decided 37 min ago. A first-time operator reads "18 h, nobody touched it". Suggest: since = time at current state after the last decision, with the stage-entry age in the hover.Needs human decisionchevron + the full Options list after a recorded hive decision; the only honest word ("stalled") is 60 chars deep in a hover. Should be a distinct state: "Hive decided adopt-supersede · stuck: workflow-path fork (can't push) · you can: Close as superseded / ask author".cue-working, alldata-motion=needs-humanor none). 4 of 6 are waiting on @clubanderson or @thatguywob; the 2 oldest hive-side rows (🐛 decision path: approve-with-rationale reads as changes_requested, Refs→Fixes rewrite overrides owner decision, 070 deadlock without a new head #14816, 🐛 [v6] queue accepts what merge will refuse: empty dashboard.authorized_users, and PAT-queued "@internal" approvals, never merge #14938) are blocked by a capability gap hive already knows about (🐛 Delegated decision on a capability gap: model the blocker, stop the un-park/re-park loop, name the missing App permission #15134, Delegated adopt-supersede reports "done" while nothing is adopted: no execution feedback, workflow-path capability gap decided as adoptable, re-adoption defeated by mutation-journal replay #15188 are the fixes) but the rows don't link to it.Screenshot: 00-full.jpg (all 39 rows).