Repository navigation
cmux-next: a new chat's effort chip names the level acpmux starts it on (Lawrence: "we lost effort") - #20577
Conversation
…on reports Lawrence 2026-10-14, "we lost effort": the New Tab composer showed the model chip alone. Red (local Chromium play on this base logic): chips = model only, waitFor did not hold. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…s it on Before a chat's session reports its options, the effort chip drew only a level a previous chat on that harness had remembered (#20335), so a fresh profile or harness showed the model chip alone. It now falls back to _acpmux/harnesses start.effort (Opus 5.5 starts at Medium). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configuration
You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…parity-surface/new-chat-effort
This comment has been minimized.
This comment has been minimized.
|
all web/size checks green |
|
Passes: CI passes on CI passes on Written by |
|
|
Gallery91 states changed: agent-pane.transcript/web-video-loaded, agent-pane.dictation/hide, agent-pane.composer/working-stop, agent-pane.composer/context-usage-menu, agent-pane.new-tab/dictation-menu, ... · 8 new states · 34 play checks failed Diff page and renders (the run's
85 more on the diff page. Live preview (follows this branch; start it with
Play checks failed (0 px anchor movement, 0 layout shift, no frame over 33 ms unless the entry gives a reason):
195 nondeterministic states differed from a second render of the same head (a clock or fixture leak) and are not counted as changes. |
|
cmux-next tour of The path-selected recorder failed for agent-pane, core; see the run artifact and failed manifests. Path-selected interactions
Agent pane (before): 3 ok
Agent pane (after): 3 ok
Core UI (before): 11 ok, 1 failed
Core UI (after): 11 ok, 1 failed
Tours are data in |
…parity-surface/new-chat-effort
![cmux-autofix[bot]](https://github.githubassets.com/images/gravatars/gravatar-user-420.png){kind=avatar}
- vp fmt webviews/test/ime-composition-keys.test.ts - vp fmt webviews/test/new-tab-omnibar-pointer.test.ts Autofix-Run: https://github.com/manaflow-ai/cmux/actions/runs/38567486409
…parity-surface/new-chat-effort
…parity-surface/new-chat-effort
…parity-surface/new-chat-effort
Bugbot needs on-demand usage enabledBugbot uses usage-based billing for this team and requires on-demand usage to be enabled. A team admin can enable on-demand usage in the Cursor dashboard. |
Conflicts with base's 10-14 composer styling: - The card, the strip and the New Tab box paint base's solid --acpmux-composer-surface; the strip takes it too so it stays one fill with the card. - The New Tab focus takes Lawrence's quiet rule (rim tone lifts a little, no accent, no wider ring) on the strip's --acpmux-card-ring. - No drop shadows stay (hq#1957 border hunt), including the permission card. - Base's pending effort chip (#20577) and mode id come in; the chips row keeps no spacer and no context ring (item 15), so base's setupTokens and RingIcon, used only by that ring, are dropped. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Problem
Lawrence (10-14): "we lost effort". With Claude Opus 5.5, the New Tab composer showed only the model chip, with no Reasoning chip.
Before a chat's session reports its options, the effort chip draws a placeholder (
PendingChip). Since #20335, that placeholder only shows a level a previous chat on the same harness remembered (chipMemory.ts). On a fresh profile, or a harness id no chat has reported on (e.g.claude-sr), it draws nothing.Change
Before the session reports, the chip falls back to the level acpmux will start the harness on:
_acpmux/harnessesstart.effort(start_defaults.rs; Opus 5.5 starts at Medium). A remembered chip still wins.Not in this PR (cx-ervhd, separate PR, waiting on SHOW LEO): the Reasoning menu itself, where a chat on the agent's own level shows "Low" with nothing checked, and Ultracode/Ultrathink sit among the levels.
Verification
new-chat-reasoningplay in local Chromium shows onlyClaude Opus 5.5among the chips, and the waitFor for "Medium" does not hold.Claude Opus 5.5 | Medium, and the play passes.test/new-chat-effort-chip.test.tsin the browser lane. Not run locally under bun test: big-red's inotify limit starves Vite's watcher, and the memory freeze bars heavy local runs.Changelog
🤖 Generated with Claude Code
Need help on this PR? Tag
@codesmith-botwith what you need.Summary by cubic
Fixes a new chat's effort chip naming the level acpmux starts the harness on, so the Reasoning chip no longer goes missing on a fresh profile or a harness no chat has reported on yet.
_acpmux/harnessesstart.effort(Opus 5.5 starts at Medium); a level remembered from a prior chat still wins.Written for commit 26540d2. Summary will update on new commits.
Showcase
In a fresh profile, open a New Tab with Claude Opus 5.5 picked. Before, from tagged DEV builds on macOS 27 at base 2834e65: no effort chip. After, at head 6305c75: a "Medium" chip next to the model. I checked that the head app's bundled pane.js contains the fix. The host chip is masked.
Full quality: 20577-before.mp4
Full quality: 20577-after.mp4
Open the live gallery