Skip to content

revert: back out crew mode toggle (#16088) - #16257

Merged
bolichen97 merged 1 commit into
mainfrom
revert/pr-16088-crew-mode
Oct 2, 2026
Merged

bolichen97 merged 1 commit into
mainfrom
revert/pr-16088-crew-mode

Conversation

@bolichen97

@bolichen97 bolichen97 commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Take the Crew Mode toggle back out of main so main is exactly what it was before #16088 landed, while the feature keeps living on its own branch.

PR #16088 ("feat: add crew mode toggle to chat sessions") was meant to stay on the dedicated branch feat/conductor-mode-toggle until the owner chose to land it. It was approved and merged into main (merge commit e53499fe4221df65eb847e2dbb1f010e8f7d706b) before that point, so main now ships a composer chip, a welcome page, a dashboard.default_crew_mode setting, 27 i18n keys in 12 catalogs, a conductor prompt section and a decision entry the owner had not yet signed off for release.

Why it matters

Anyone building from main today gets a user-visible feature (and a persisted setting) that is still being decided on. If it stays, the next release carries it, the i18n catalogs and the settings registry freeze around it, and taking it out later means a real removal with migration work instead of a clean revert while nothing downstream depends on it yet.

Not a goal

  • Judging the feature itself. Nothing in feat: add crew mode toggle to chat sessions #16088 is being rejected on its merits; it comes back through a fresh PR from feat/conductor-mode-toggle when the owner is ready.
  • Touching anything else that landed on main after the merge. The revert is git revert -m 1 of the one merge commit; the 22 files other PRs changed in the meantime keep those changes.
  • Deleting or rewriting history. Main keeps the merge commit; this adds one commit on top.

What changed (motivation → approach → change)

Symptom: main carries #16088 ahead of the owner's decision. Cause: the PR was approved and merged by another maintainer while it was meant to stay parked on its branch. Change: one commit, git revert -m 1 e53499fe4221df65eb847e2dbb1f010e8f7d706b, which applied cleanly with no conflicts, plus three deliberate departures from a byte-exact revert, each forced by a gate or a recorded decision (all caught by CI rounds 1 and 2):

  • docs/decisions/2026-10-02-conductor-introduces-itself-in-every-chat.md and its row in docs/decisions/README.md stay. The decision ledger is append-only (decisions/README.md rule 2, enforced by the Decision Ledger Gate): an entry is never deleted, and an agent may not write a superseding one. The decision Bolin recorded there still stands, and it binds every Conductor chat, not only Crew Mode chats.
  • Because of that entry, the "Talking to the person" section of the kirocrew-conductor prompt in src/kiro_crew/agent.py stays, byte-identical to what feat: add crew mode toggle to chat sessions #16088 merged, together with the two spec-digest pins that lock it in (test/test_agent_refactor_contracts.py, test/test_agent_refactor_spec_bytes.py). The section's one mention of the Crew Mode switch is an example of who may be reading, not a dependency on it; it is left as the maintainer approved it so the pins stay unchanged. Main therefore keeps doing what the decision entry says. (The first head of this PR deleted the section; the GPT and First Principles lanes both blocked on recorded-decisions-are-not-regressions, correctly.)
  • The one-word brand fix feat: add crew mode toggle to chat sessions #16088 made in docs/system-specs/modules/learn-cron-dashboard.md ("KiroCrew logo" -> "Kiro Crew logo") is kept, because reverting it re-adds a misspelling the Brand Name Gate rejects.

Net: 42 files, 22 insertions, 1229 deletions, the mirror of the merge minus those three items.

What goes away, all of it added by #16088:

  • Backend: DashboardConfig.default_crew_mode and its loader line; the default_crew_mode read/validate/echo branches in api_dashboard_config; the chat.turn-on-crew-mode-for-new-sessions-by-default row in settings-registry.generated.json. The conductor prompt section stays (see above).
  • Frontend: CrewModeChip, CrewModeWelcome, lib/crewMode.ts, the chip and welcome wiring in ChatPage.tsx, the newChatAgent hook-ins in chat-sidebar/create.ts, shellKeyboard.ts, paletteActions.ts, actionsProvider.ts, the Settings → Chat → Sessions row in ChatPanel.tsx, the generated settingsRegistry.gen.ts entry, the glossary.json term and the 27 keys in every locale catalog including en-XA.
  • Docs: the Crew Mode row in docs/feature-map/README.md, and the paragraphs added to session.md, learn-cron-dashboard.md, config.md, crew-mode.md, agents.md, work-ledger.md. The decision entry stays (see above).

Nothing left on main references the removed code: a grep for crewMode, CrewMode, crew_mode, default_crew_mode, CREW_MODE across src/ and website/src/ (locales included) returns nothing; the only remaining mention of conductor-introduces under docs/ is the kept decision entry and its README row. The pre-existing crew-mode.md spec and its "Retired: Crew Mode" references are about the older retired slot mode and are untouched by both the merge and this revert.

The feature branch was restored first: feat/conductor-mode-toggle is back on this repository at the PR's head, b4551ba8b95bba0633db210a27dc758fdfbff9d1, with no PR open from it.

Backwards compatibility

Breaking: the optional dashboard.default_crew_mode setting, on main for a few hours, is removed again. Writer sweep on origin/main @ af162847d6: the only writers were the Settings → Chat → Sessions switch (ChatPanel.tsx) and api_dashboard_config itself, both removed here; nothing else on main, in cron, subagent, app or release code, writes it.

Reader: src/kiro_crew/config/loader.py:_build_dashboard_config -- chat -- a default_crew_mode value a user already saved in config.yaml is kept as an unknown in-section key and ignored, the same path every unmodelled key takes; nothing loads as an error.
Reader: src/kiro_crew/dashboard/handlers/files.py:api_dashboard_config -- chat -- a stale, not-yet-reloaded dashboard tab that still PUTs default_crew_mode gets the existing Unknown fields 400 and nothing is written; a reload clears it. test/test_dashboard_files_coverage.py keeps pinning the Unknown fields 400.
Reader: website/src/pages/chat-sidebar/create.ts:createSlot -- chat -- new chats start on the default agent again, exactly as before the merge; ChatPage.memoryChip.test.tsx is restored to its pre-merge assertion.

Tests

No tests are added; the ones #16088 added go away with the code they covered:

  • test/test_crew_mode.py (deleted) covered the setting default, PUT/GET round-trip and the prompt section.
  • test/test_agent_refactor_spec_bytes.py and test/test_agent_refactor_contracts.py keep the digests and prompt fingerprint feat: add crew mode toggle to chat sessions #16088 pinned for kirocrew-conductor / kirocrew-ledger-conductor, which lock in that the kept prompt section is byte-for-byte what the maintainer approved (74 tests, pass locally).
  • website/src/lib/crewMode.test.ts, CrewModeChip.test.tsx, ChatPage.crewMode.test.tsx (deleted); ChatPage.memoryChip.test.tsx back to asserting the chip row leaves once the session has messages.

Local gate on this worktree: scripts/check_black_formatting.py (pinned black, fresh cache) pass, check_subprocess_encoding.py pass, isort --check-only pass, flake8 clean, mypy --platform linux src/kiro_crew clean (2062 files), scripts/docs_lint.py pass, check_decisions_history.py and check_brand_name.py against origin/main pass, cd website && npm run build pass; scripts/local-gate.py and npm run test results are posted in a PR comment as they finish.

Manual verification

N/A — this is a mechanical revert of one merge commit with no conflicts; the identifier grep above and the spec-byte pins are the check that nothing dangles.

Screenshots / video

Why no screenshot: the diff removes the Crew Mode chip, welcome page and settings row, taking the chat composer and Settings → Chat → Sessions back to exactly the surfaces main rendered before #16088; there is no new rendered state to show.

Related Issues

Reverts #16088. The feature continues on feat/conductor-mode-toggle.

Pattern harvest

Not generalizable: a process slip, a parked PR got approved and merged by a second maintainer; the fix is the merge-hold convention on the PR, not a code rule.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

@bolichen97
bolichen97 requested a review from a team October 2, 2026 20:57
@bolichen97
bolichen97 requested a review from a team as a code owner October 2, 2026 20:58
@bolichen97
bolichen97 requested a review from CrysisDeu October 2, 2026 20:58
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 2, 2026
@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Fable 5) — ✅ PASS

UX-level review of 9ec43cc37e40f37990cee6032063778ec5ecef69 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: PASS

Nothing to check.

[UX-REVIEWED] 9ec43cc

Description read

[DESCRIPTION-READ] f6a5dc24c327ea0651347853bc0b133ca32326bdbb0cc2d8e198398a931ce380

That sha256 names everything this verdict read: the PR description, plus the 3 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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Fable 5) — ✅ PASS

Design-level review of 9ec43cc37e40f37990cee6032063778ec5ecef69 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Confirm feat/conductor-mode-toggle really holds the feature at b4551ba8b before merging — this revert assumes that branch is the surviving copy.

[DESIGN-REVIEWED] 9ec43cc

Description read

[DESCRIPTION-READ] 7d7fe5e8d5c008f03b23be5915117f39adf2015a12e4f1d160b06ceb949c1210

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of 9ec43cc37e40f37990cee6032063778ec5ecef69 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 9ec43cc

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 9ec43cc37e40f37990cee6032063778ec5ecef69: <one-sentence reason>

@bolichen97
bolichen97 force-pushed the revert/pr-16088-crew-mode branch from 92b2189 to 7a58ce1 Compare October 2, 2026 21:03
@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Fable 5) — ✅ human override accepted

@bolichen97 overrode this lane for 9ec43cc37e40f37990cee6032063778ec5ecef69 via /ai-review override (this commit only).

@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Oct 2, 2026
PR #16088 was merged into main before the owner meant to land it. This
reverts merge commit e53499f so main
no longer carries the crew mode toggle; everything else that landed on
main since stays intact.

The feature itself is not lost: it stays on feat/conductor-mode-toggle
(restored at b4551ba, the PR head) and
will come back to main through a fresh PR when it is ready.

This reverts commit e53499f.
@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5 Review — ✅ no blocking findings

Reviewed 9ec43cc37e40f37990cee6032063778ec5ecef69 — this comment is updated in place on each push.

Review details

Nothing blocks; one advisory — the conductor's shipped system prompt still names the dashboard switch this revert removes.

FINDING — src/kiro_crew/agent.py:3951 — the "Talking to the person" section added by #16088 and kept here still tells every kirocrew-conductor turn the person "may have turned on Crew Mode, the dashboard switch that runs a chat on you", a control this PR deletes from the composer, Settings, the config schema and the PUT allowlist, so the conductor reasons from and can name a switch no longer in the UI (and the test that pinned the string is deleted in the same diff) → Fix: drop the they may have turned on **Crew Mode**, the dashboard switch that runs a chat on you clause from _CONDUCTOR_SYSTEM_PROMPT, keeping the rest of the section the recorded decision docs/decisions/2026-10-02-conductor-introduces-itself-in-every-chat.md requires — an edit to a file this PR does not otherwise touch, hence advisory.

[OPUS-REVIEWED] 9ec43cc

Verdict parsed from the review's SHA-scoped output markers for commit 9ec43cc37e40f37990cee6032063778ec5ecef69.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 9ec43cc37e40f37990cee6032063778ec5ecef69: <one-sentence reason>

@bolichen97
bolichen97 force-pushed the revert/pr-16088-crew-mode branch from 7a58ce1 to 9ec43cc Compare October 2, 2026 21:26
@bolichen97

Copy link
Copy Markdown
Contributor Author

Round-2 dispositions, applied in 9ec43cc:

  • GPT 5.6 F1 (src/kiro_crew/agent.py, removing the Conductor introduction reverses docs/decisions/2026-10-02-conductor-introduces-itself-in-every-chat.md): real, fixed. The "Talking to the person" prompt section and the two spec-digest pins (test/test_agent_refactor_contracts.py, test/test_agent_refactor_spec_bytes.py) are kept exactly as feat: add crew mode toggle to chat sessions #16088 merged them. git diff origin/main...HEAD -- src/kiro_crew/agent.py is now empty.
  • First Principles item 2 (same section): fixed by the same change, the subtraction the lane proposed.
  • First Principles item 1 / Blockers (removing the Crew Mode toggle is a product-shape change with no RFC or recorded decision): not something this PR can clear on its own. The authority is the repository owner's instruction to back the toggle out of main until he lands it himself; it is not on record in docs/ and an agent may not record it. This needs a maintainer's /ai-review override first-principles 9ec43cc37e40f37990cee6032063778ec5ecef69: <reason> on this PR, or stays red.
  • Frontend Tests (2), hoverVariantPolicy.test.ts "DECORATIVE entry still matches a site: components/WelcomeView.tsx": not from this diff. Both files are byte-identical to origin/main; the same test fails on a clean main tree (the sm:group-hover:line-clamp-none snippet left WelcomeView.tsx in fix(chat): keep floating composer chrome off the transcript and hero #15820 after the policy list was written). Fix PRs already open: fix(website): repair hover policy and SessionActionsMenu tests after #14807 #16223 and fix: repair two frontend tests broken by the hover-gate cross-merge #16210. Not rerunning the shard until one of them lands.
  • Build Windows Installer (x64): hit the 120-second install ceiling in test-windows-installer.ps1. The diff touches no installer, wheel or dependency file; watching whether round 3 reproduces it before calling it anything.

@bolichen97

Copy link
Copy Markdown
Contributor Author

Round 3 on 9ec43cc: GPT 5.6 cleared (no findings), Screenshot Evidence, Brand Name Gate, Decision Ledger Gate and Build Windows Installer (x64) are green. Three reds remain, none from this diff:

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 2, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author

/ai-review override first-principles 9ec43cc: owner backed the Crew Mode toggle out of main until he lands it himself; the feature stays on feat/conductor-mode-toggle

@github-actions

github-actions Bot commented Oct 2, 2026

Copy link
Copy Markdown
Contributor

Human judgment recorded

@bolichen97 marked the first-principles AI finding as false positive, not applicable, or explicitly accepted for 9ec43cc37e40f37990cee6032063778ec5ecef69.

owner backed the Crew Mode toggle out of main until he lands it himself; the feature stays on feat/conductor-mode-toggle

This decision applies only to this commit. A new push requires a new judgment.

@bolichen97

Copy link
Copy Markdown
Contributor Author

/ai-review override first-principles 9ec43cc: Owner decision: #16088 was merged before I meant to land it. Back Crew Mode out of main; it stays on feat/conductor-mode-toggle until I land it myself.

🤖 AI-assisted comment 🤖

@bolichen97
bolichen97 merged commit a9cf34a into main Oct 2, 2026
92 of 98 checks passed
@bolichen97
bolichen97 deleted the revert/pr-16088-crew-mode branch October 2, 2026 22:06
@github-actions

github-actions Bot commented Oct 2, 2026

Copy link
Copy Markdown
Contributor

Human judgment recorded

@bolichen97 marked the first-principles AI finding as false positive, not applicable, or explicitly accepted for 9ec43cc37e40f37990cee6032063778ec5ecef69.

Owner decision: #16088 was merged before I meant to land it. Back Crew Mode out of main; it stays on feat/conductor-mode-toggle until I land it myself.

This decision applies only to this commit. A new push requires a new judgment.

@github-actions

github-actions Bot commented Oct 2, 2026

Copy link
Copy Markdown
Contributor

Human judgment recorded, but the following reviewer lane(s) could not be re-run automatically: First Principles Review. Re-run the lane's latest workflow run for 9ec43cc37e40f37990cee6032063778ec5ecef69 manually from the Actions tab.

iamwhatever pushed a commit that referenced this pull request Oct 17, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. An empty chat with it on shows the Crew Mode welcome page.
The switch is not offered when the conductor itself is the default
agent, since off would have nothing to switch back to. There is no new
setting, route, config key or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: Bolin Chen <17506219+bolichen97@users.noreply.github.com>
iamwhatever pushed a commit that referenced this pull request Oct 17, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. An empty chat with it on shows the Crew Mode welcome page.
The switch is not offered when the conductor itself is the default
agent, since off would have nothing to switch back to. There is no new
setting, route, config key or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: Bolin Chen <17506219+bolichen97@users.noreply.github.com>
iamwhatever pushed a commit that referenced this pull request Oct 17, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. An empty chat with it on shows the Crew Mode welcome page.
The switch is not offered when the conductor itself is the default
agent, since off would have nothing to switch back to. There is no new
setting, route, config key or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: Bolin Chen <17506219+bolichen97@users.noreply.github.com>
iamwhatever pushed a commit that referenced this pull request Oct 17, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. An empty chat with it on shows the Crew Mode welcome page.
The switch is not offered when the conductor itself is the default
agent, since off would have nothing to switch back to. There is no new
setting, route, config key or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: Bolin Chen <17506219+bolichen97@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
bolichen97 added a commit that referenced this pull request Oct 18, 2026
A Crew Mode chip right of "Choose memory mode" switches a plain chat to
the existing kirocrew-conductor through the ordinary agent switch, and
back again. Turning it on in a root session opens the side panel
Dashboard view, whose Overview is that session's own Dynamic Dashboard.
Turning it off leaves the side panel and the page alone. A dispatched or
adopted session switches without opening it.

On the server, the switch to the conductor template on a root session
adopts the shipped goal-board on that session's page with the ordinary
instance.adopt, only while nothing is adopted there yet, and tells an
open Overview to re-read. Each dashboard turn of a root session with an
adopted page carries its own [DASHBOARD] block (the crewmate block keyed
by the session's store key), so the conductor's existing rule skips its
inline task board.

The empty-chat welcome page keeps its title, tagline, four capability
lines and ghost art, and drops the sample board: one line and an "Open
task board" button point at the side panel instead.

The chip keeps one accessible name ("Crew Mode") with aria-pressed
carrying on and off, stays focusable while a reply runs, and its busy
and on hints show as a visible line it names with aria-describedby.

The client root test (isRootSession) is the side panel's existing one,
moved into command-center/model.ts so the panel and the switch share it.
There is no new setting, route, agent, template or session field.

Re-lands the switch backed out in #16257 on the current page owners.

Co-authored-by: zejiangg <zejiangg@amazon.com>
Co-authored-by: Joe Guo <12429733+iamwhatever@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant