Repository navigation
fix(chat): show a plain name for the Conductor where a chat's agent is labelled - #22944
Conversation
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: a display-only label change in one shared helper, keyed on the kind each session recorded, with rendered evidence of this revision. Notes
[DESIGN-REVIEWED] 0e1c9af Description read[DESCRIPTION-READ] 7a1590d35a133cff3c2c08a76fa89cd1543615a12daf75857eef87da52c4737c 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 blocks. Issue #21531 is the source of the relabel, and the new tests fail on base. The one leftover is a lookup built to hold more names than the one it carries. Notes
What this change shipsInventory (8 items) — 7 justifiedIntent: someone using Crew Mode sees "Conductor" instead of the code name
[FIRST-PRINCIPLES-REVIEWED] 0e1c9af Description read[DESCRIPTION-READ] 9f35224ed5e0cda51690a25f99f7d8ebdbeb8733aae5b6baefe381715a525496 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 screenshot shows the new "Conductor" label on the session row and on the composer chip, and the blind reader read it correctly as "the name of the helper or assistant that ran this chat". Notes
[UX-REVIEWED] 0e1c9af Description read[DESCRIPTION-READ] 9d7f1e0bc1c1e77584cb3ca2f3514212e166f10652a0c2ed0a493785993f93f5 That sha256 names everything this verdict read: the PR description, plus the 4 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: |
bolichen97
left a comment
There was a problem hiding this comment.
Reviewed head 0cfe1b2. The change looks good to ship. I found no blocking or should-fix issues.
What I checked:
agentOrDefaultLabelmaps the name only when the slot recordedagent_kind === 'template'. A same-named crewmate and a kind-less pin keep the raw id. The old'default'special case still behaves the same, becausebuiltinAgentDisplayName('default')returns'default'.- Every chip and row that labels a live slot now passes the slot's kind. History rows and schedule jobs keep the raw name, as the description says. The raw id still drives tint, routing and search; only the display string changes.
- The new key exists in all 13 checked-in catalogs, and
en.manual.jsoncarries the English source. - No file is on a security-sensitive path. All five AI review lanes pass on this head.
Remaining maintainer checks before approval: the readiness label is still checking, PR Readiness is pending, and Backend Tests (Windows) shard 7 is red on this head. This PR changes frontend files only, so that shard is probably unrelated, but it must be green or rerun first. I will look again when readiness settles.
|
Status: the only red check on head 0cfe1b2 is 🤖 AI-assisted comment 🤖 |
0cfe1b2 to
0e1c9af
Compare
|
Review-ready on head 0e1c9af. 🤖 AI-assisted comment 🤖 |
bolichen97
left a comment
There was a problem hiding this comment.
Approving head 0e1c9af.
Gates checked on this head:
- Readiness label is passed and PR Readiness is green.
- Required checks are green, including Coverage Gate.
- All AI review lanes name this head, including UX. None of them blocks.
- No open change request. The linked issue is tier T2. No security path is touched.
This head is a rebase only. The range diff matches the commit I reviewed before, so my earlier review stands. The Conductor now shows a plain name wherever a chat's agent is labelled.
Problem / Motivation
Goal: A chat that runs on the built-in Conductor shows "Conductor" as its agent label, not the code name
kirocrew-conductor.In a chat on the
kirocrew-conductortemplate (Crew Mode), the session row and the composer chip showkirocrew-conductor. A user who just turned Crew Mode on cannot tell whether that is another assistant. Deferred from the UX review of #21526.Why it matters
Every Crew Mode chat shows the code name on its row and under its message box. It reads as jargon and makes "same chat, one switch" look like a different assistant.
Not a goal
What changed (motivation → approach → change)
agentOrDefaultLabelis the one label every surface uses for a chat's agent, and it only knew the name. It now also takes the kind the session recorded (agent_kind). Atemplatepick ofkirocrew-conductorreads "Conductor"; an agent-less chat whose default template is the Conductor reads "Conductor · default". A crewmate with the same name, a kind-less pin, and every other agent keep the raw name.The session row (
SessionRow.tsx), composer chip (ComposerDockView.tsx), chat pane (ChatPane.tsx), agent sync hook (useAgentSync.ts) and crew window (crewWindowAgentMode.tsx) now pass the slot's kind. History rows and schedule jobs record no kind, so their pins stay raw. The raw id still drives tint, routing and search. "Conductor" is a newcomponents.agentLabel.builtin_conductorkey in every catalog underwebsite/src/i18n/locales/(en-XAregenerated byi18n:pseudo), reusing each locale's existing word for conductor.Backwards compatibility
Compatible: the new
kindargument is optional, and every label without the Conductor template reads exactly as before.Removes nothing: only a display string changes; no id, key, field or path is taken away.
Tests
website/src/utils/agentLabel.test.ts(new): the template conductor reads "Conductor"; a same-named crewmate and a kind-less pin keepkirocrew-conductor; an unknown agent stays raw; the inherited default reads "Conductor · default"; French reads "Conducteur". Onorigin/main4 of 6 fail (the crewmate and kind-less cases already pass there and guard against a name-only mapping).website/src/test/ChatSidebar.defaultAgentMarker.test.tsx: a rendered session row shows "Conductor" for the template slot and the name for the crewmate slot. Fails whenSessionRowstops passings.agent_kind.--sequence.shuffleseeds 1-5, 17/17 each run.run_scoped_tests.py --surface frontend): 82 files, 1323 tests pass.src/i18n/50 files pass;tsc -p tsconfig.app.json, eslint andlint:i18nclean.Manual verification
Ran the built SPA with stubbed slots (
website/scripts/capture-21531-conductor-label.mjs): the template conductor's row and composer chip read "Conductor", the same-named crewmate row readskirocrew-conductor, akirocrewrow is unchanged.Screenshots / video
Related Issues
Closes #21531
Pattern harvest
Not generalizable: one label helper gained a display name for one built-in template; there is no repeated defect shape to rule out.
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)