Repository navigation
fix(dashboard): show a worker-hosted discussion on the conductor mode Board - #39263
Conversation
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the change adds a missing UI reader for a field the server already sends, follows the goal-board rule, removes nothing, and its rendered evidence matches the diff. Notes
[DESIGN-REVIEWED] d67dffa Description read[DESCRIPTION-READ] 429a3a1bd732b136ec051680dbb565ef50ece42c25cf47134a47397ee33d7c71 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 |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing to check. Every item fixes the reported defect (#39227): the Board tab shows a discussing milestone as "In progress" because What this change shipsInventory (9 items) — 9 justifiedIntent: let a person see on the default Board tab that a worker-hosted discussion is running, and what it has agreed so far. FIX.
[FIRST-PRINCIPLES-REVIEWED] d67dffa Description read[DESCRIPTION-READ] 188ffc25d3ea29efaff581e412bfea2b7c4fa7f70426b29fab3b4e8ee39565ed That sha256 names the PR description this verdict read. Recompute it with |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing blocks. The primary toggle ("Show/Hide the discussion") was read correctly with confidence, and every new state appears in a screenshot. Notes
[UX-REVIEWED] d67dffa Description read[DESCRIPTION-READ] cec2a3be7898933653f051d1d4a65114140226bef29cbc37d0cf5d69a6d42cc2 That sha256 names everything this verdict read: the PR description, plus the 7 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 |
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 findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
c7cdfde to
d67dffa
Compare
|
Rebased: main moved in test/test_kas_token_refresh_recovery.py (the red bare-session-handle guard, fixed by #39262) and website/src/i18n/locales/* since head c7cdfde (c5ac14c -> 6e059d7). Same push fixes the hi/bn register of this PR's own new strings (tum/tumi forms) flagged by hiStyle/bnStyle. CI re-running. |
Code change approver recommendation: A person should decideWhy: the decider estimates a 10% chance that this change introduces a defect, within the 12% limit for approving on its own, but it changes 446 lines, more than the 400 it may approve on its own. This adds a read-only "discussing" state to the conductor mode Board, using the Dashboard tab template's own rule ( Tier: pending: being reviewed. 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.
Approved. On this head the review checked these points. The new discussing state uses the same rule, in the same order, as the goal-board template, and needs-you still wins. A closed, orphaned or stale item still falls through as before. Every discussion field renders as React text. The thread link is drawn only when it matches the server's own pattern and is under the length cap, and a test proves a javascript: link draws nothing. The new pytest pins that /api/crew-board sends discussion. The milestoneState tests cover both sides of every guard. PR Readiness is green.
Done well: the Board tab and the Dashboard tab now agree, and there is no server change.
Optional:
ConductorModeCrewPanel.tsx:319: a worker's finaldonereport that still carries adiscussionnow reads "Discussing" instead of Done until the conductor accepts it. You could skip the discussing branch whenitem.status === 'done'.ConductorModeCrewPanel.tsx:858: the recap counts a discussion as running even when its host is idle. The body says this happens only while the worker is mid-turn. Neither the recap nor the pulse at line 833 has a test, so you could gate the recap onalive === 'running'(or reword the body) and pin it.src/kiro_crew/docs/work-ledger.md:215still names only the goal board as a reader ofdiscussion.
Problem / Motivation
Goal: The conductor mode Board side panel shows a worker-hosted discussion as "Discussing", with its round, minutes and counts, and a panel that opens to show what the discussion holds.
When a worker hosts a discussion, the Board tab shows that milestone as plain "In progress". The person cannot tell a discussion is running, which round it is on, or what has been agreed. #37340 taught only the goal-board template (the Dashboard tab) to draw this row; the Board tab is a different component that never reads
discussion.Why it matters
The Board is the tab a conductor chat opens by default, so it is where people look. Every worker-hosted discussion today looks like ordinary work there, and the agreed / disputed points are only visible by opening the worker's chat.
Not a goal
AssistantMessage.tsxand the card component): another PR owns it./api/crew-boardalready sendsdiscussion(a test here pins it).What changed (motivation → approach → change)
Symptom: a discussing milestone reads "In progress". Cause:
ConductorModeCrewPanelhas no discussing state andWorkBoardItemhas nodiscussionfield, though the board payload carries it.milestoneStategainsdiscussing, placed after needs-you and before running, with goal-board's own rule: item open,discussion.phaseset, worker not closed, not orphaned, not stale. A closed item or a stale worker falls back to its plain state.WorkBoardItem.discussionis typed inwebsite/src/api/crewBoard.ts. New copy in all 13 catalogs underwebsite/src/i18n/locales/(en-XA.jsonregenerated), worded as goal-board's.website/capture/conductor-mode-discussing.tsxrenders the real panel with fixtures.Backwards compatibility
Compatible: the new field is optional; an item without
discussionkeeps every state it had.Removes nothing: the diff only adds a state, a field, copy and tests.
Tests
ConductorModeCrewPanel.test.ts: discussing for a running or idle host; needs-you still wins; accepted / abandoned close it; stale → quiet, closed or orphaned → stuck; no phase → no discussion.ConductorModeCrewPanel.discussion.test.tsx: the row text, counts, toggle and every panel section render; an unsafe thread link is not drawn; a closed item and a stale worker show no discussion.test/test_work_ledger_board.py::test_a_hosted_discussion_reaches_the_board_row:/api/crew-boardcarries the report'sdiscussion.Proof: not provable -- frontend test (prove.py runs Python only)
Manual verification
Rendered the capture page in Chromium (dark, light, zh-CN, panel closed and open); screenshots below.
Real-machine verification
N/A -- a dashboard render change with no host or install path; unit tests and the capture page cover it.
Screenshots / video
Panel open (dark):
Closed, zh-CN, light
Related Issues
Closes #39227
Sibling sweep
Swept:
git grep -n "discussion" -- website/srcSibling: src/kiro_crew/dashboard_templates/builtin/goal-board/template.html -- not affected -- already draws the row since #37340
Pattern harvest
Not generalizable: one server field gained a second UI reader that was never written; the field itself is generic.
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...); PR Hygiene does not count commits