Repository navigation
fix(dashboard): fold dropped milestones into one row on the conductor mode Board - #39201
Conversation
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: it is the same fold the done rows already use, it is reversible, and the screenshots committed in this revision show it. Notes
[DESIGN-REVIEWED] 48dec1b Description read[DESCRIPTION-READ] 03dd9d4f69daa28df65bf37eace588d79528cc6f5a3fb0e8316501a9367e83c6 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing blocks. The blind reader identified the new "N dropped" fold in both its folded and opened states, said what a click does and would use it. The fold is the same expander as the existing "N done" row, so nothing is swapped out. Notes
[UX-REVIEWED] 48dec1b Description read[DESCRIPTION-READ] fdaaed39a5de975f804ee3ea85c82cb8d55ed3aa7a5c70ec6e0d4537a397fc0d That sha256 names everything this verdict read: the PR description, plus the 9 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing blocks. The fold reverses the spec's "stays among the open rows" pin, and issue #39192 supports that with a user report of about 40 dropped rows making the Board unusable. Not checked: the workflow patch spans about 180 files from main's history; I reviewed the PR's 25-file change against main. Notes
What this change shipsInventory (9 items) — 8 justifiedIntent: keep live milestones visible on the conductor mode Board after the conductor drops many. This is a FIX.
[FIRST-PRINCIPLES-REVIEWED] 48dec1b Description read[DESCRIPTION-READ] 0ef483f060c230e93afbf761a00977c4f56baa153a1fc34eab42bf4c22b96f2a That sha256 names the PR description this verdict read. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
GPT 6.1 Review — ✅ no blocking findings (all downgraded on adjudication)GPT 6.1 flagged blocking issues on This comment is updated in place on each push. Review detailsBLOCKING -- website/src/components/SharedCrewsCards.tsx:107 -- Keep failed Crew opens on the error surface
Adjudication (Opus 5.5) — is blocking on each finding proportionate?F1 — LOW harm, and blocking this PR prevents none of it. Both lines the finding names, False positive or not applicable? A repository writer can comment: |
aa1a71f to
fb175a6
Compare
|
Rebased: main moved in docs/system-specs/modules/crew-mode.md (and other files under docs/system-specs/modules/) since this head went green (1171295 -> be0b414); CI re-running. The new head fb175a6 also takes both First Principles notes: the "When a run is done" spec clause now says an abandoned milestone sits in the "N dropped" fold, and the done and dropped folds share one local |
fb175a6 to
c83df06
Compare
c83df06 to
2478982
Compare
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Thanks for this. The code and the new test look right on this head. One evidence item blocks the merge.
website/src/pages/chat/command-center/ConductorModeCrewPanel.tsx:756-757, 783, 889: the PR changes the Status view. Abandoned milestones leave the open list, and a new "N dropped"<details>row now sits below the open rows. The PR has no image of this. The Design and UX reviews on head 2478982 both say they cannot evaluate it without a screenshot, and the UX blind read did not run. A review that cannot reach a verdict because evidence is missing counts as a block. Smallest fix: no code change is needed. The capture pagewebsite/capture/conductor-mode-run-states.tsx?state=done-somealready mounts the real panel with one abandoned milestone, so it renders a "1 dropped" fold now. Capture the Status view on main (before) and on this head (after), with the fold both closed and open, at the default width. Dark and light would help. Attach the images to the PR body or commit them undertemp-screenshots/, then re-run the Design and UX lanes on that head.
What the review checked on this head: only abandoned items are folded, and a rejected milestone stays in the open list. The header count, Recap and the stuck, quiet and moving checks still read the full lists. The done fold keeps its classes and test id through the shared milestoneFold helper. The new test pins what goes in the fold, its position and its closed default. The head also merges cleanly with current main.
Optional:
CrewWorkerList.test.tsx:615: the/0\D+4/check matches the text of the whole milestones section, not just the heading. Scoping it to theh3with an exactMilestones · 0 of 4would make the pin tighter.- Nothing toggles the dropped fold open. A toggle step like the one in the done-fold test would catch a mix-up of the two folds' open state.
2478982 to
ea5b8d7
Compare
Code change approver recommendation: A person should decideWhy: the decider estimates a 7% chance that this change introduces a defect, within the 12% limit for approving on its own, but it is held at tier 2: changes requested. On the Board's Status view, this PR moves milestones closed as abandoned out of the open list and into a folded "N dropped" row below it. Rejected milestones stay in the open list, and the milestone count, Recap and run line are unchanged. Hiding a dropped row still works inside the fold, and the "N done" fold is unchanged now that both folds share one helper. The spec and all twelve languages are updated. Two new tests check the fold and its placement, the remove test was updated to the new row order, and CI is green. Nothing blocks. Tier: 2, held for a person: changes requested. An automated recommendation, not a review: a maintainer still decides. It approves on its own only when the estimated chance of a defect is at most 12% and the change has at most 400 changed lines. Decider v3c (217fe607) · commit |
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Thanks for the rebase. The code still looks right on this head, but the evidence item from the earlier review is still open.
website/src/pages/chat/command-center/ConductorModeCrewPanel.tsx:1108: abandoned milestones now leave the open list (line 972) and go into a new "N dropped"<details>row below the open rows, so the Status view looks different. The PR body still has no image, and the diff adds no file undertemp-screenshots/. Because of that, the Design and UX reviews on ea5b8d7 both say they cannot evaluate the change, and the UX blind read did not run. When a review cannot reach a verdict because evidence is missing, that counts as a block. No code change is needed to fix this.website/capture/conductor-mode-run-states.tsx?state=done-somealready mounts the panel with one abandoned item, so it renders a "1 dropped" fold. Capture the Status view on main (before) and on this head (after), with the fold both closed and open, at the default width. Dark and light would help. Attach the images to the PR body (or commit them undertemp-screenshots/), then re-run the Design and UX lanes on that head.
Earlier review on 2478982:
- Missing before and after screenshots: still stands. The rebase added no image, and there is no reply to the review.
What the review checked on this head:
- Only
abandoneditems fold. A rejected milestone stays in the open list. - The header count, the Recap and the stuck, quiet and moving checks still read the full lists.
- The done fold keeps its classes and test id through the shared
milestoneFoldhelper. - The updated remove test re-pins the moved Hide target and the new row order without loosening any assertion.
Done well: one helper now draws both folds, with no copied <details> block, so the two folds cannot drift apart.
Optional:
CrewWorkerList.test.tsx:615: the/0\D+4/check matches text anywhere in the milestones section. An exactMilestones · 0 of 4check on the heading would pin the count more tightly.- No test toggles the dropped fold open. A toggle step like the one in the done-fold test would catch a mix-up between the two folds' open state.
- When a settled run has only abandoned milestones left, the empty open list puts the done fold's bottom border right on top of the dropped fold's top border, so the divider looks doubled.
…ts screenshots Pin the Milestones heading text exactly, toggle the dropped fold open without opening the done fold, and drop the dropped fold's top rule when only the done fold sits above it, so the divider is not doubled. Add a component capture for the fold and its screenshots.
|
Addressed in 48dec1b:
|
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Approving head 48dec1b. The review checked these points on this head:
- Only
abandonedmilestones move into the new "N dropped" fold. Rejected ones stay in the open list. - The header count, the Recap and the stuck, quiet and moving checks still read the full milestone set.
- Hide still works for rows inside the fold, and focus falls back to "Show again".
- The new key is in all locale files.
- The crew-mode spec line is updated in the same PR.
Earlier review on ea5b8d7, item by item:
- Missing screenshots: fixed. Six renders are committed under
temp-screenshots/, and the Design and UX lanes both pass on this head. - Heading count check: fixed. The test now checks the exact
Milestones · 0 of 4heading. - Dropped fold toggle: fixed. A new test opens the dropped fold and checks that the done fold stays closed.
- Doubled divider: fixed. The dropped fold drops its top rule when it sits right under the done fold.
Done well: one milestoneFold helper now draws both folds, so their markup and toggle wiring cannot drift apart.
Optional:
website/capture/conductor-mode-needs-input.tsx:19,24still importsPREVIEW_CREW_MODE, which main no longer exports. That is the same fix as the run-states entry, and you could add it here or send it as a small follow-up.CrewWorkerList.test.tsx:595: anexpect(fold).toHaveClass('border-t')in the test with open rows would pin the divider above the fold.- The screenshots could move to PR attachments with
gh pr edit --attach, so they do not stay in main's history.
Problem / Motivation
Goal: On the conductor mode Board, dropped milestones fold into one "N dropped" row below the open ones, so the live work stays on screen when the conductor changes course.
A user reported a run where the conductor changed course and dropped about 40 milestones. The Board's Status view folds done milestones into one "N done" row, but each dropped one stays as a full "Not done · dropped" row among the live ones. The Board became unusable and the user restarted the session to get a clean one.
Why it matters
The Board is where the owner checks what is running and what needs them. A long goal that changes plan collects many dropped milestones, and each one looks like work that has to be read and skipped. The goal board Dashboard page already folds abandoned tasks (#38300); the Board tab should match.
Not a goal
What changed (motivation → approach → change)
Dropped milestones were never split out of the open list: the Status view folded only
donerows, and everything else, closed ones included, was drawn as a row.ConductorModeCrewPanel.tsxnow takes the milestones closed asabandonedout of that list and draws them under one<details>row, "N dropped", below the open rows. The row is folded by default and opens to the same milestone rows as before. It works the same way as the existing "N done" fold. With no dropped milestone there is no row.The new label
conductor_mode_ms_dropped_foldis added in all twelve languages underwebsite/src/i18n/locales/(en.manual.jsonand the eleven translations, plus the regenerateden-XA.json).docs/system-specs/modules/crew-mode.mdsays how the fold behaves.Backwards compatibility
Compatible: the panel reads the same board fields; only where dropped rows are drawn changes.
Reader: website/src/pages/chat/command-center/ConductorModeCrewPanel.tsx:ConductorModeCrewPanel -- crew page -- dropped rows are hidden behind one click, not removed;
Milestones: dropped milestones fold into one "N dropped" row below the open oneschecks they are all inside the foldTests
CrewWorkerList.test.tsx›Milestones: dropped milestones fold into one "N dropped" row below the open ones: two dropped milestones sit in a folded "2 dropped" row, the running and the rejected one stay in the open list above it, and the header still counts all four. It failed on main (no fold row) and passes with the fix. Turning the fold back into the open list makes it fail again.CrewWorkerList.test.tsx›Milestones: the dropped fold opens on its own, and drops its top rule when only the done fold sits above it: toggling the dropped fold leaves the done fold closed, and the dropped fold has no top rule when no open row separates the two folds. Forcing the rule back on makes it fail.npx vitest run src/pages/chat/command-center/ src/i18n/: all pass (247 and 1288 tests).Proof: not provable -- frontend test (prove.py runs Python only)
Manual verification
tsc,eslinton the changed files,i18n:checkagainst main and semgrep on the diff all pass.Real-machine verification
N/A -- a dashboard rendering change; the component test drives the real panel.
Screenshots / video
Component render, not the live page. These come from
website/capture/shoot-conductor-mode-dropped-fold.mjs, which mounts the realConductorModeCrewPanelwith mocked board data at side-panel width. To retake them:cd website && node capture/shoot-conductor-mode-dropped-fold.mjs.Before (main): every row from the dropped fold sat in the open list, mixed in creation order with the running and failed rows.
Related Issues
Closes #39192
Sibling sweep
Swept: git grep -n "abandoned" -- website/src/pages/chat/command-center/ src/kiro_crew/dashboard_templates/builtin/goal-board/
Sibling: src/kiro_crew/dashboard_templates/builtin/goal-board/template.html -- not affected -- abandoned tasks already fold there since #38300
Sibling: website/src/pages/chat/command-center/ConductorModeCrewPanel.tsx -- changed here
Pattern harvest
Not generalizable: one list on one panel that folded one closed state and not the other.
Checklist