[teamleaderleo]

Leo: "if you could just, hot reload and test out the f***ing fix right then and there, without much fanfare at all ... and it's still like the same kind of flow, right?"

Leo-ask: ask-ffdcb76e34

What

scripts/cmux-next/try-pr.sh <PR> is one command for trying a PR. It reads the PR's diff with git (no GitHub API), sorts it with scripts/cmux-next/live_apply.py, and takes the cheapest path:

• web-live: runs web-overlay.sh follow <PR>, so the running app hot-reloads the PR's web UI in seconds with no build.
• no-app (CI, tests, docs) and ios-only: reports that there is nothing to try in the Mac app.
• daemon, swift-live, rebuild: says why it needs a new binary (a stored property, a changed signature, a new file, a resource), and with --build requests the same fleet dev build as today, tagged pr-<N>.

--dry-run prints the verdict and the command without running it. The rules come from the live-apply study of 8,315 merged PRs: 66% of NEXT-era PRs that touch the app needed no new binary, and most of those are web changes. The hot-reload doc gets a row and a section.

Not in this PR: injecting Swift into the app itself. The view shell has live code today (#35314), but the app is not linked for it yet. For a body-only Swift change the script says it is injectable later and needs a dev build for now.

Verification

• python3 -m unittest scripts/cmux-next/tests/test_live_apply.py: all 10 tests pass, including a real git range.
• try-pr.sh N --dry-run --build on merged PRs: Add the view shell and cmux preview open: the UI in real windows without the app #35314 → rebuild (new files, Package.swift), test(cmux-tui): the returning phone test waits until the daemon saw the phone leave #35408 → no-app, Webview agent pane waits on live daemon requests #35206 and ui: compact shared tooltip previews #35205 → web-live (web-overlay.sh follow).
• prepush-gate.py: no new violations.

Observed / Not run

Not run: web-overlay.sh follow on a Mac against a live app. That path is the existing, unchanged script. A real --build submission was not run either.

Changelog

none

🤖 Generated with Claude Code
Agent session: agent session

---

^{Open in [code]smith · Need help on this PR? Tag @codesmith-bot with what you need.}

---

Summary by cubic

Adds scripts/cmux-next/try-pr.sh <PR> and live_apply.py for trying a pull request without waiting for a build when none is needed. The script reads the PR's diff with git, classifies it with live_apply.py, and takes the cheapest path: web-only changes go straight to web-overlay.sh follow for live hot-reload, CI/tests/docs changes report nothing to try, and daemon, Swift, or resource changes say why they need a new binary (with --build requesting the fleet dev build tagged pr-<N>). --dry-run prints the verdict and command without running.

Addresses review follow-ups: lockfiles, bundled scripts, and web config (like vite.config.mjs) no longer fall into no-app—the extension rule now runs after the web, resource, and Rust rules. The hot-reload doc gives honest numbers (about half of recent NEXT-era PRs need no new binary) and stops counting swift-live as needing no build. --port with no value prints usage. All 10 tests pass, including dry runs on several merged PRs.

^{Written for commit 89ae0b8. Summary will update on new commits.}

[github-actions]

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

Review 89ae0b8: HQ feed card · gallery before/after runs · checks

App of this head (app.zip, kept 1 day)

gh workflow run reload-build.yml -R manaflow-ai/cmux --ref feat-cmux-next -f tag=pr35450 -f ref=lane/try-pr-live -f platform=macos -f backend_mode=local

red: cmux-next checks (god files, concurrency, crash safety, l10n); no finished base run on 08052d3 to compare; do not merge until green; see lane-rules

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.4184409870901 (78ee0e108e)
• cmux NEXT nightly 1.0.0-nightly.4184683079001 (64822b9434)
• cmux NEXT nightly 1.0.0-nightly.4184857252001 (8c9aae8b0d)
• cmux NEXT nightly 1.0.0-nightly.4185261809301 (697838c74d)
• cmux NEXT nightly 1.0.0-nightly.4185464793401 (68821b04eb)
• cmux NEXT nightly 1.0.0-nightly.4185545977501 (b659812803)
• cmux NEXT nightly 1.0.0-nightly.4185901282301 (b8d9ef9cab)
• cmux NEXT nightly 1.0.0-nightly.4185979991201 (8a134ee405)
• cmux NEXT nightly 1.0.0-nightly.4186744784501 (890be4f45a)
• cmux NEXT nightly 1.0.0-nightly.4186855704101 (53421cdccd)
• cmux NEXT nightly 1.0.0-nightly.4187086980301 (df4454e051)
• cmux NEXT nightly 1.0.0-nightly.4188039004101 (3e464b08ff)
• cmux NEXT nightly 1.0.0-nightly.4190094575701 (036f8aa538)
• cmux NEXT nightly 1.0.0-nightly.4191131679101 (e01bbae5c9)

[teamleaderleo]

Review (subagent claude-opus-5.5 medium): SHIP at 89ae0b8

Checked: bash on 3.2 (empty arrays under set -u, quoting), the fetch and merge-base range, the classifier order after the fix, and that the docs match the behaviour. Unit tests and dry runs pass.

Follow-ups: webviews/*.md and non-Rust tooling under cmux-tui/ get a heavier verdict than they should (safe side); use the PR's own base rather than feat-cmux-next; take the merge-base from FETCH_HEAD; Swift rule misses (multi-line signatures, nested stored properties, didSet, tabs) before swift-live drives injection.

[blacksmith-sh]

Found 10 test failures on Blacksmith runners:

Failures

Test	View Logs
/tmp/tmp534aqc5t/summary-fail.json (live):/summary-fail.json (live):	View Logs
/tmp/tmpulnwynfv/summary-missing.json (live):/summary-missing.json (live):	View Logs
[new_workspaces] debug.new_workspace timed 0 of 50 new workspace rows/ [new_workspaces] debug.new_workspace timed 0 of 50 new workspace rows	View Logs
[new_workspaces] the window draws no frames/[new_workspaces] the window draws no frames	View Logs
[tabs] debug.tab_switch timed 0 of 40 switches/ [tabs] debug.tab_switch timed 0 of 40 switches	View Logs
[tabs] the window draws no frames/[tabs] the window draws no frames	View Logs
imported web bundle inputs are not routed:/imported web bundle inputs are not routed:	View Logs
nightly cmux-tui resolve/nightly cmux-tui resolve	View Logs
nightly.yml must not read vars.NIGHTLY_SIGN while nightly-sign.yml is in shadow mode/ nightly.yml must not read vars.NIGHTLY_SIGN while nightly-sign.yml is in shadow mode	View Logs
the short nightly decide gate must use the pooled Linux route with a Blacksmith fallbac k/ the short nightly decide gate must use the pooled Linux route with a Blacksmith fallbac k	View Logs

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

[teamleaderleo]

Review (codex, router-review): SHIP at 89ae0b8

Reviewer: codex gpt-6-astra high, fast tier.

Follow-ups: 1. They land as checklist items in this PR's one follow-up tracking issue in this repo (Follow-ups: #N ..., label follow-up; add to an open one before filing, never one issue per finding), linking this receipt; the lane that opened the PR fixes them forward.

SHIP at 89ae0b8

Visual verdict: no images; Leo tries the build

• [P2] [follow-up] [category:correctness] [finding:F1] [outcome:pending] scripts/cmux-next/live_apply.py:35: REBUILD_FILE only matches /Resources/, so root bundled paths such as Resources/foo.sh fall through to NOAPP_EXT and are classified no-app. The app bundles root Resources/; those changes should request a rebuild.

Focused unit tests, syntax checks, and diff checks pass.

Follow-ups:

• [finding:F1] Include root-level Resources/ in resource detection.

[cmux-bot]

cmux-bot approves 89ae0b8: an independent review passed this exact head (router-review receipt by teamleaderleo). A push needs a new review; a later BLOCK at this head dismisses this approval.