[teamleaderleo]

Leo finds the side padding in agent chats too wide and uneven, and sees the chat jump and jitter while a reply streams.

What changes

• The transcript and the composer share one readable column. A narrow pane leaves the page edge (--page-edge, 18px, from CmuxNext: one container level and consistent edge padding (batch 1) #20121) on each side. A wide pane widens the column up to a readable cap and keeps it centered.
• The view stays at a streaming reply's latest line unless you scroll up. Nothing moves when a queue appears, when a row opens, or when a reply streams a few words at a time.

Before

The transcript's rows sat in a 720px column inside 26.5px gutters. The composer sat in a 760px column inside 16px gutters, so their edges never lined up.

Proof

webviews/test/agent-pane-chat-column.test.ts (Chromium) measures the column at four pane widths. It also reads every painted frame while a reply streams, a queue appears, a command opens and a reply streams word by word. The test commit is red on base; the results are below.

Changelog

• Changed: Agent chats read in one column shared by the transcript and composer, with page-edge side padding, and stay pinned to the bottom while a reply streams.

🤖 Generated with Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need.}

---

Summary by cubic

The agent chat now reads in one column shared by the transcript and composer, stays pinned to the streaming reply's latest line, and styles prompts as solid bubbles with a single Copy action.

Previously the transcript used a 720px column inside 26.5px gutters and the composer a 760px column inside 16px gutters, so their edges never lined up. Content growth only glided on the first placement step, so streamed text dropped by the estimate's error and slid past its place at each token; now every step glides and the moves add up. Queued prompts lay over the transcript's foot, and a prompt was a translucent, unevenly padded bubble with two Copy buttons on hover.

• The column is 720px, 62.5% of a wider pane up to an 800px cap, centered, at least the page edge (18px) from each side; model.ts and the minimap mirror it.
• The transcript now ends above queued prompts, and rows reveal a floating Copy only for keyboard focus and a selected range (a prompt's hover shows the one under its bubble, 12px text insets, 10px radius).
• Adds a Chromium regression test (agent-pane-chat-column.test.ts) that measures the column at four pane widths and samples every painted frame while a reply streams, a queue appears, and a command opens. Followed and code-card rest positions allow a few pixels of settle, since line heights are fractional.

^{Written for commit 9709c3b. Summary will update on new commits.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: Repository: manaflow-ai/cmux/.coderabbit.yaml
• Review profile: ASSERTIVE
• Plan: Advanced
• Run ID: 4836b1e6-ec57-41f2-96f5-8aed99ed4f88

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[cursor]

Bugbot is paused — on-demand spend limit reached

Bugbot uses usage-based billing for this team and has hit its on-demand spend limit.

A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue.

[blacksmith-sh]

Found 1 test failure on Blacksmith runners:

Failure

Test	View Logs
nightly cmux-tui resolve/nightly cmux-tui resolve	View Logs

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[github-actions]

all web/size checks green

[github-actions]

Passes: CI passes on 9709c3b39c.

CI passes on 9709c3b39c (run 38507224073 attempt 1).

Written by scripts/ci/classify_failures.py (ci-failure-attribution.yml); signatures are its SIGNATURES table. A machine verdict is the runner's fault, not this PR's; yours means the failing file is one this PR changes, also red on main that main's latest full suite fails the same way, seen on other PRs that it failed on another pull request's run lately.

[github-actions]

CI fast guards passes on 9709c3b39c (https://github.com/manaflow-ai/cmux/actions/runs/38507224073). This covers only the fast guards, not CI: CI's result is the ci-status check, and the CI failure attribution comment names any failing test.