[teamleaderleo]

HOLD: lands after #26791, whose fleet package tests have not run yet (the fleet controller is stalled).

What

Where a new chat runs becomes one choice in the composer: This Mac or Cloud. Behind Labs (preview features).

• The control. A Cloud row at the top of the environment picker's Cloud section, with one line on what it does. Every existing row stays: the account's machines, New Cloud Machine, Open Cloud…, SSH… and the rest. Picking it makes the composer read Cloud; a machine picked after it makes the chat a chat on that machine again.
• The default stays This Mac. Nothing flips it.
• The first send with Cloud picked calls cloud.startChat (cmux NEXT: cloud.startChat, a new chat's first send runs in Cloud #26791) with the prompt, the folder the folder chip names, the agent and the model. The host picks and sizes a Cloud machine, copies the folder, opens the chat there and closes this tab. No machine, size or environment prompt.
• Refusals keep the prompt. Attachments are refused on Cloud for now. A Cloud pick that is no longer available when the send happens (Labs went off, a chat started here) is refused too, never sent as a chat on this Mac.
• One start at a time. While a Cloud start waits on the host (a new machine boots in minutes), another Return or New chat starts nothing.
• Picking Cloud drops a saved environment or machine picked before it, so the folder sent is the one the chip shows. It also hands focus back to the prompt, so Return sends.
• The row lives in the existing picker, so it opens the way that picker already does; this PR adds no menu or animation.

Gallery

Gallery: agent-pane.cloud-run-designs/row, agent-pane.cloud-run-designs/bare, agent-pane.cloud-run-designs/chip, agent-pane.composer-context/environment-picker-cloud-picked

Three designs side by side in agent-pane.cloud-run-designs (CloudRunDesigns.tsx, experimental). The two alternatives are gallery-only and wrap the real scope row; only the picked one ships.

• row (picked): Cloud is a row at the top of the picker's Cloud section, with one line on what it does.
• bare: the same row with no subtitle.
• chip: a This Mac / Cloud switch beside the picker.

Why row: one control already answers where a chat runs, and the subtitle tells Cloud apart from a named machine.

States of the shipped row stay in agent-pane.composer-context: environment-picker-cloud-row, -cloud-picked, -cloud-search and -cloud-closed.

Why

Leo, 2026-10-28: "You can just say ... where should this chat run and stuff, and that's just about it ... it's somewhere in the cloud, and then it gets managed for you. You don't have to select an environment, you don't have to create an environment, you don't have to do all that nonsense".

Risk

• Behind Labs. Without it the row does not show and the host method refuses with labs_off.
• The environment picker's other rows are unchanged; persistent() now resets any non-persistent mode (Sandboxed or Cloud) when a machine is picked.
• The model goes to the host, but the machine's start uses the agent's default model for now (cmux NEXT: cloud.startChat, a new chat's first send runs in Cloud #26791).
• A folder-history default (a folder whose last chat ran in Cloud opens with Cloud picked) is not in this PR.

Tests

• EnvironmentPicker.test.tsx (new): picking Cloud sets the mode and focuses the prompt on the first pick and on a repeat pick; a machine picked while Cloud is the mode goes back to a chat here.
• NewTabScreen.test.tsx: a Cloud send goes to onCloud, not a new chat, and a refusal keeps the prompt; a saved environment picked then Cloud sends the chip's folder (/repo/A), not the tab's inherited home.
• cloudRun.test.ts: the host call's params, its refusals, attachments refused, a Cloud pick that is no longer offered refused, and a second start while one waits refused.
• Linux webview checks (bun run check) and these suites ran with cmux-ci test at this head.
• test/gallery-coverage.test.ts fails the same two tests on feat-cmux-next (a2f8b65), with the same list of components; none is in this PR. A base red.

Review: an independent Codex (gpt-6-astra, high) review took three rounds and returned SHIP at 8098947; a delta review of the gallery designs and the one-start guard returned SHIP at ca765c9. Since then only the gallery-only switch design changed (a fieldset for lint).

Stacked on #26791. It is retargeted to feat-cmux-next once that lands.

Captures: the Gallery: line above picks the states the gallery advisory run renders. Per #21400 its images stay in the run's artifact (gallery-advisory-<PR>-<HEAD_SHA>), not on this public PR.

Agent session: agent session

🤖 Generated with Claude Code

[teamleaderleo]

Owned by an agent session · session

[teamleaderleo]

Review (codex gpt-6-astra): SHIP at ca765c9

[cursor]

Bugbot needs on-demand usage enabled

Bugbot 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.

[blacksmith-sh]

Found 168 test failures on Blacksmith runners:

Failures

Test	View Logs
src/pages/settings/layout.test.ts/layout.test	View Logs
src/pages/settings/layout.test.ts/layout.test	View Logs
test/accent-glass-invariants.test.ts/accent-glass-invariants.test	View Logs
test/accent-glass-invariants.test.ts/accent-glass-invariants.test	View Logs
test/agent-pane-agents-pill.test.ts/agent-pane-agents-pill.test	View Logs
test/agent-pane-agents-pill.test.ts/agent-pane-agents-pill.test	View Logs
test/agent-pane-background-tasks.test.ts/agent-pane-background-tasks.test	View Logs
test/agent-pane-background-tasks.test.ts/agent-pane-background-tasks.test	View Logs
test/agent-pane-chat-unavailable.test.ts/agent-pane-chat-unavailable.test	View Logs
test/agent-pane-chat-unavailable.test.ts/agent-pane-chat-unavailable.test	View Logs

...and 158 more test failures. View all on Blacksmith

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

[cmux-bot]

cmux status _{(one comment, edited in place; each line is updated by the job that owns it)}

Unclear: web / react-apps-check failed: no known signature; failed step: Check webview test shards.

CI failed on d1838f3216 (run 40659638104 attempt 1): 4 code, 1 unknown.

Job	Verdict	Why
web / react-apps-base-check (2)	code	a web test failed
web / react-apps-base-check (3)	code	a web test failed
web / react-apps-base-check (4)	code	a web test failed
web / react-apps-base-check (1)	code	a web test failed
web / react-apps-check	unknown	no known signature; failed step: Check webview test shards

Failures:

• seen on other PRs layout.test.ts only page.css imports the page's sheets: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts chromium > a prompt the fold hides, run elsewhere, leaves no room as the stack shrinks, opens and cl...: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts webkit > the queue opened, the view scrolled to its end shows the last row whole, and it stays: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts webkit > the queue closed at the latest row keeps the last row on the stack, with no room that glide...: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts webkit > a prompt the fold hides, run elsewhere, leaves no room as the stack shrinks, opens and clos...: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts webkit > a prompt past the opened queue's cap, run elsewhere, leaves no room when the queue closes: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts webkit > in a chat switched to, a queued prompt withdrawn leaves no room held for the last chat's qu...: failed (job)
• seen on other PRs agent-pane-queue-scroll-bottom.test.ts webkit > a new line in the composer leaves the transcript's distance from its end as it was: failed (job)
• seen on other PRs agent-pane-dock-gutters.test.ts chromium > the gutter beside every part of the stack scrolls the transcript, and the part keeps the ...: failed (job)
• seen on other PRs agent-pane-dock-gutters.test.ts webkit > the gutter beside every part of the stack scrolls the transcript, and the part keeps the po...: failed (job)
• ... 22 more in the job logs

Matched log lines

web / react-apps-base-check (2): (fail) only page.css imports the page's sheets [36.16ms]
web / react-apps-base-check (3): (fail) chromium > the gutter beside every part of the stack scrolls the transcript, and the part keeps the pointer [8549.77ms]
web / react-apps-base-check (4): (fail) the summary card lists the chat's background tasks: running ones first, the finished folded [1168.90ms]
web / react-apps-base-check (1): (fail) no orange or system accent and no opaque glass on the gallery stages [489614.74ms]

Not re-run automatically: web / react-apps-base-check (2), web / react-apps-base-check (3), web / react-apps-base-check (4), web / react-apps-base-check (1), web / react-apps-check are not machine failures.

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.

CI fast guards failed on d1838f3216 (https://github.com/manaflow-ai/cmux/actions/runs/40659638104). It does not block the merge; a red guard merged into main breaks it for every open PR.
The log named no failed step; see the run.

Agents: python3 scripts/ci/guard_attribution.py fix applies the mechanical fixes locally.

[teamleaderleo]

Folded into #28686 (original commits kept, on current feat-cmux-next). Review and landing continue there; the owner closes this one.

[teamleaderleo]

Owned by an agent session · session · took over from an earlier session (its owning session is closed)

[teamleaderleo]

Closing as a duplicate: #28686 folds this PR (with #26801, #26920 and #26921) into one PR on feat-cmux-next and carries it from here. This branch stays on the remote.