Skip to content

fix(chat): say what Switch All Sessions keeps and what it stops - #18843

Merged
cruisercohen merged 1 commit into
mainfrom
fix/11231-switch-all-copy
Oct 10, 2026
Merged

cruisercohen merged 1 commit into
mainfrom
fix/11231-switch-all-copy

Conversation

@cruisercohen

@cruisercohen cruisercohen commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: The Switch All Sessions dialog tells the user what a switch really does: the conversation is kept, and only a running reply that is not skipped gets stopped.

The dialog says, in red, "Switching a session resets its conversation". Most people read that as "my messages are gone". They are not. A bulk switch restarts the session's agent process, and the conversation comes back on the next message.

Why it matters

Everyone who opens the dialog sees the warning. An overstated warning either scares people off a safe action, or teaches them to ignore red text in the one case where it matters: a session that is mid-reply and not skipped really does lose that reply.

Not a goal

  • Naming which sessions are affected, skipped or failed instead of just counting them. That is the second slice of the issue. It needs a separate UX decision and a payload change, and is left for a later PR.
  • The other count fixes in the issue (labelling the button's count, saying what a retry after a failure will switch).
  • Any change to what a switch does on the backend.

What changed (motivation → approach → change)

Symptom: the warning says a switch "resets its conversation", but on main nothing in the switch path deletes the conversation.

What a switch does, read from main's src/kiro_crew/dashboard/chat_handlers.py and src/kiro_crew/session_lifecycle.py:

  • Transcript: kept. The reset keeps the persistence entry.
  • Agent context: kept. The reset keeps the session's resume id, so the next turn reloads the native session (session/load). If that load fails, or the provider changes, the conversation is replayed from the log instead. Either way the conversation continues on the new model.
  • Queued messages: kept. They are parked and handed to the restarted session.
  • Running turns: skipped by default. If the user clears "Skip N running sessions", the in-flight reply is stopped. That is the only thing a switch loses.
  • Sessions with sub-agents: always skipped.

Change: the copy now says this. In website/src/pages/ChatSidebar.tsx the description reads "Pick a model for every session. The conversation is kept and continues on the new model." The red caveat, "Running sessions switch too, and their current reply is stopped.", shows only when sessions are running and the user has cleared the Skip box, because that is the only case where anything is lost. With the default (Skip ticked), there is no red text. The old two-fragment pair (one sentence split across two keys) is replaced by two complete sentences, so translators work with whole sentences.

website/src/i18n holds the catalogs. All 12 authored locales are translated, each in its style guide's register (du / tú / tu / ты / तुम / তুমি / 你, です・ます, 합니다체). The old fragments had several of these wrong (for example German "Sie", French "vous"). en-XA is regenerated with npm run i18n:pseudo, and en.context.json gets translator context for both keys. npm run gen:ui produced no change to ui-index.generated.json, because this text is not indexed.

Rejected alternatives:

  • Drop the warning. That would also drop the one real caveat: an unskipped running reply gets cut off.
  • List the affected sessions by name now. It is a bigger change, and it is its own UX and payload decision. It is deferred, as noted above.

Backwards compatibility

Compatible: copy only. No API, setting or behaviour changes.

Two catalog keys are removed and replaced (pages.chatSidebar.pick_a_model_for_every_session_switching_a_sessi, pages.chatSidebar.resets_its_conversation):

Reader: website/src/pages/ChatSidebar.tsx:ChatSidebar -- chat -- the only reader on origin/main; it now reads the replacement keys (ChatSidebarCoverage.test.tsx)

Tests

  • website/src/test/ChatSidebarCoverage.test.tsx: the panel says the conversation is kept, and the old "resets its conversation" text is gone. The red caveat is absent while Skip is ticked and appears once it is cleared. It is also absent when no session is running.
  • src/i18n suites (parity, QA on changed values, per-locale style tests) are green with I18N_BASE_REF=origin/main. lint:i18n, i18n:check, check-i18n-keys, tsc and eslint also pass. The related frontend set passes too: 146 files, 2213 tests.

Manual verification

Built the SPA and opened the panel against stubbed APIs in Playwright, with one running session: English with Skip ticked (default), English with Skip cleared, and German with Skip cleared. The capture asserts that the caveat is present or absent in each case. Text wraps cleanly at sidebar width.

Screenshots / video

Default, Skip ticked (nothing at risk, no caveat):

Switch All Sessions, skip ticked

Skip cleared (running sessions will switch, caveat shown):

Switch All Sessions, skip cleared

German, Skip cleared

Switch All Sessions in German

Before: "Pick a model for every session. Switching a session resets its conversation." (red, always shown).

Related Issues

Part of #11231. Naming the affected sessions is deferred to a follow-up slice.

Open PR #13619 also touches this dialog (it adds effort picking) and keeps the old reset copy. Whichever lands second rebases.

Pattern harvest

Rule candidate: review-prompt
Pattern: destructive-sounding UI warning copy that is not checked against what the handler actually destroys

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

@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 10, 2026
@github-actions

github-actions Bot commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 6b793788b445f7f17d34e914f13ee863eed19044 — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: PASS

Check that the conversation really survives a bulk reset on every selectable harness, not just kiro-cli. The new copy promises this to everyone who opens the dialog.

What this change ships

Inventory (6 items) — 6 justified

Intent: stop the Switch All Sessions dialog from saying a bulk model switch wipes the conversation, and warn only when a running reply will really be cut off. This is a FIX.

  1. The description line under the heading now says the conversation is kept and continues on the new model, instead of "resets its conversation" — justified
  2. The red warning is hidden by default and appears only when sessions are running and the Skip box is cleared — justified
  3. The red warning now says the current reply of a running session is stopped — justified
  4. Two sentence-fragment catalog keys are replaced by two complete-sentence keys, translated in all 12 authored locales plus the en-XA pseudo-locale — justified
  5. The German, French, Hindi, Bengali and Russian strings now use the informal register — justified
  6. The dialog's code comment and the translator context now describe a restart that keeps the conversation — justified

[FIRST-PRINCIPLES-REVIEWED] 6b79378

Description read

[DESCRIPTION-READ] 980fa0a05b897194f93942501c9ec13e7ccb345fe31f348901467ff57c2ec5f3

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ✅ PASS

UX-level review of 6b793788b445f7f17d34e914f13ee863eed19044 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: PASS

Check that "their current reply is stopped" makes clear the half-written reply is lost. The blind reader could not tell whether it was lost or kept.

Suggestions

  • bulk_switch_stops_running_reply: say what is lost in plain words, e.g. "Running sessions switch too. The reply in progress is stopped and lost." The PR says this is the one thing a switch destroys, so the caveat should state it directly.

[UX-REVIEWED] 6b79378

Description read

[DESCRIPTION-READ] 75789277b69edb326fa97247e79d149f084ce3ec8deb6cb43b6e0dd60f5f5a2b

That sha256 names everything this verdict read: the PR description, plus the 6 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 10, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 6b793788b445f7f17d34e914f13ee863eed19044 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Nothing to check.

[DESIGN-REVIEWED] 6b79378

Description read

[DESCRIPTION-READ] a4c5a07c3e8ff9a894b46211262a1048d5a3aa261e83dc9d0b44119275de4447

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 10, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 6b793788b445f7f17d34e914f13ee863eed19044 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 6b79378

Verdict parsed from the review's SHA-scoped output markers for commit 6b793788b445f7f17d34e914f13ee863eed19044.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 6b793788b445f7f17d34e914f13ee863eed19044: <one-sentence reason>

@github-actions

github-actions Bot commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 6b793788b445f7f17d34e914f13ee863eed19044 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 6b79378

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 6b793788b445f7f17d34e914f13ee863eed19044: <one-sentence reason>

@cruisercohen
cruisercohen marked this pull request as ready for review October 10, 2026 12:18
@cruisercohen
cruisercohen requested review from a team and KiroCrewCaptain as code owners October 10, 2026 12:18
@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 10, 2026
@cruisercohen
cruisercohen force-pushed the fix/11231-switch-all-copy branch from 3386705 to 6b79378 Compare October 10, 2026 12:36
@cruisercohen

cruisercohen commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=bc9c2f1a624d The red caveat shows by default even when nothing is at risk.

fixed in 6b79378: the caveat now renders only on running sessions AND the Skip box cleared, so the default state shows no red text.
Copy rewritten as a direct consequence: "Running sessions switch too, and their current reply is stopped." (no double negative), in all 12 locales.
Test: ChatSidebarCoverage asserts absent while Skip is ticked, present after clearing it.

@cruisercohen

cruisercohen commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=cf2b4d29a001 "Restarts each session's agent" reads as jargon.

fixed in 6b79378: the sentence is dropped in every locale; the description now reads "Pick a model for every session. The conversation is kept and continues on the new model."

@cruisercohen

cruisercohen commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=d5f5a314bbd0 German checkbox label joined by word order.

rebutted (out of goal): the "Skip N running sessions" label is two pre-existing keys (skip + running_session) that this PR does not touch.
Merging them into one interpolated string is a catalog restructure across 12 locales and pluralKeys, separate from correcting the reset warning; it belongs with the issue's deferred count/label slice.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: passed Eligible automated validation passed for the current revision labels Oct 10, 2026
@cruisercohen

Copy link
Copy Markdown
Contributor Author
  • span=1e693ce014ef Say what is lost in plain words.

rebutted: "stopped" is the claim the code supports. A force switch tears the turn down; whether the partial text already streamed survives in the transcript is not something this PR verified, so "lost" could overstate it the way "resets its conversation" did.
The caveat already names the loss directly (the current reply) and shows only in the one case where it applies.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention labels Oct 10, 2026
@cruisercohen
cruisercohen enabled auto-merge (squash) October 10, 2026 13:07

@iamwhatever iamwhatever left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This replaces the red "Switching a session resets its conversation" warning in the Switch All Sessions dialog with copy that matches what a switch does: the conversation is kept, and only an unskipped running reply is stopped. The fix is at the root. The one misleading sentence, which was split across two keys, is now two whole-sentence keys in all 12 locales plus en-XA and en.context.json. The red caveat renders only when bulkRunningCount > 0 and the skip box is cleared, which is the only case where anything is lost (ChatSidebar.tsx:4685). The new ChatSidebarCoverage tests check the kept-conversation text and that the old text is gone. They also check the caveat is absent by default and when nothing is running, and present once Skip is cleared, so they would fail on main. All checks are green, base is main, and all three design lanes plus GPT and Opus pass. Every UX item from earlier heads has a disposition, and the current-head "stopped vs lost" note is rebutted with a reasonable not-verified argument. Copy-only change with no behaviour or API change. — crew-pr-reviewer

@cruisercohen
cruisercohen merged commit 5904af8 into main Oct 10, 2026
102 checks passed
@cruisercohen
cruisercohen deleted the fix/11231-switch-all-copy branch October 10, 2026 19:08
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 10, 2026
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.

2 participants