[teamleaderleo]

Leo (2026-10-29): "well somehow at "100% cpu" usage claude code desktop still seems fast / how does that work? how can cmux next also do that? let's do that"

and: "but i also wonder how much of this is simply putting the desktop at higher priority?" ... "the gui of the desktop app i mean"

This is the app side of the QoS work; the process side is #27587.

Where the main thread stalls under load

The measurements in #27587 (scripts/measure/qos-load-bench.sh, cmux-mac-mini, 10 cores) answer Leo's question:

• CPU is not the problem. A thread at the main thread's QoS runs every 120 Hz frame on time at 100% CPU (0 late of 21,600). cmux already runs at 46, like Claude Desktop's 47-54, so raising the GUI's priority gains nothing.
• Disk on the main thread is the problem. Under agent-like disk load (git, rg, find, dd), a 64 KiB Data.write(options: .atomic) save took 1.0 s at p95, and 44% of them took longer than a frame, whatever priority the load ran at. Uncached reads and fsyncs stall too (p99 86 ms; fsync p50 33 ms).

An audit of the main thread (every file under Packages/macOS/CmuxNext/Sources) found few such waits: daemon socket I/O already runs on actors, and blocking calls go through checked wrappers. Ranked by how often they hit the interactions Leo named:

main-thread wait	trigger	how often	this PR
AgentUnsentDrafts.save: atomic write of every tab's draft	typing in a new chat	every settled edit (300 ms throttle)	moved off main
AgentPaneScrollAnchors.save: atomic write of up to 500 places	scrolling a transcript	every settled scroll	moved off main
AgentPaneScrollAnchors first read	switching to or opening a chat	once per run	still a read, now ordered behind queued saves
AgentPaneBrowserChoices.list: LaunchServices queries, Info.plist reads, icon renders for 16 apps	opening a preview card's "Open in" menu	each menu open	next
RemoteRdStreamTransport queue.sync per key and pointer event	remote browser input	each event	next
TerminalOutputLane queue.sync on close	closing a terminal tab	each close	next
icon(forFile:) in drag and Open With menus	file drags, menus	each	next

What this does

AgentPaneStateFile (new, CmuxNextAgentPane) writes one small JSON state file on its own serial utility queue:

• Saves keep call order. A save that a newer one replaced before it ran is skipped.
• JSON encoding runs on the queue too, so the main thread only hands over a value.
• read() waits for the saves already queued for that file, so a store made right after a save reads it back. Each path has one writer.
• Quit (applicationWillTerminate and the forced-quit path) waits for queued saves under the reviewed allowBlocking(.terminate) allowance.

The unsent-drafts store and the scroll-place store use it. No new timer or observer: a queue runs only when a save is queued, and quit drains it.

Performance

metric	before	after	delta	harness or run link
main-thread time per draft save, under agent disk load, p95	1.0 s	one enqueue (no disk on main)	-100% of the disk wait	scripts/measure/qos-load-bench.sh (#27587), atomic_write_us
main-thread time per draft save, idle, p50	0.8 ms	one enqueue	-100% of the disk wait	same
saves that block the main thread longer than a frame, under load	44% (132 / 300)	0	-100%	same

The "after" column holds by construction (the main thread never touches the file). App-level input-to-frame numbers for typing and scrolling under load are not here: they need the GUI bench on a capture mini (the every-action sweep in scripts/measure/blocking-probe.sh under synthetic load), which is the next step.

Testing

• AgentPaneStateFileTests: a save runs off the main thread and a read sees it, a burst ends with the last value, nil removes the file, flushAll waits, and unsent drafts and scroll places survive a fresh store.
• UnsentDraftTests and ChatScrollRestoreDefaultPathTests (existing) still read back right after a write.
• scripts/cmux-next/check-concurrency.sh: ok.

No UI changes, so there are no screenshots.

No capture: nothing on screen changes; two small state-file saves move from the main thread to a background queue.

Agent session: agent session

🤖 Generated with Claude Code

---

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

---

Summary by cubic

Moves the agent panes' state file saves off the main thread so typing and scrolling no longer stall on disk I/O under agent load.

• AgentUnsentDrafts and AgentPaneScrollAnchors now save through a new AgentPaneStateFile, which encodes and atomically writes on one serial utility queue per file, keeps call order, and skips saves superseded before they run. A throwing encode leaves the file as the last save that ran left it; nil removes it.
• Reads wait for the file's queued saves; only a read from inside that file's own save reads the disk directly, and each file queue has its own on-queue key so a save can't read another file's saves out of order. App quit flushes pending saves under the terminate allowance, bounded by a shared deadline that reports whether all drained.
• Under agent disk load the old main-thread save took 1.0 s at p95 and 44% of saves ran over a frame; the main thread now only enqueues the save.

^{Written for commit 828c4f0. Summary will update on new commits.}

[teamleaderleo]

Owned by an agent session · session

[blacksmith-sh]

Found 9 test failures on Blacksmith runners:

Failures

Test	View Logs
--from must install a complete prebuilt tree/ --from must install a complete prebuilt tree	View Logs
/tmp/tmp0qb6_o5h/summary-fail.json (live):/summary-fail.json (live):	View Logs
/tmp/tmpfckji1pb/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
test_every_listed_hook_is_one_glaeda_released_with_the_check (main.SigningHooksInSt epWithGlaedaTests.test_every_listed_hook_is_one_glaeda_released_with_the_check)/ test_every_listed_hook_is_one_glaeda_released_with_the_check (main.SigningHooksInSt epWithGlaedaTests.test_every_listed_hook_is_one_glaeda_released_with_the_check)	View Logs
test_the_repository_highlights_validate (main.DigestTests.test_the_repository_highl ights_validate)/ test_the_repository_highlights_validate (main.DigestTests.test_the_repository_highl ights_validate)	View Logs

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

[github-actions]

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

red: cmux-next checks (god files, concurrency, crash safety, l10n); no finished base run on b6554ad to compare; do not merge until green; see lane-rules

[teamleaderleo]

Review (Codex gpt-6-astra high, fast, review router): SHIP at be44791

The review took four rounds:

• 712f0b3: BLOCK. flushAll had no bound. read() could deadlock on the file's own queue. An encode failure deleted the file. The tests never forced overlap.
• baba105: BLOCK. One shared on-queue key let a save of one file read another file directly. Two tests could race.
• 6f4c7ac: BLOCK. The test's queue hold could expire before the check that needed it.
• be44791: SHIP, no findings.

The fixes:

• flushAll waits on every file queue against one shared deadline and reports whether it drained.
• Each file queue has its own DispatchSpecificKey, so a read never waits on its own queue.
• A throwing encode keeps the file as the last completed save left it, and nil still removes it.
• The tests hold the queue so saves really overlap.
• The crash window (the queue backlog) is written down in the type's docs.

828c4f0 is a clean merge of the current base, which now includes the JoinConsent compile fix #27758. Nothing was resolved by hand.

Fleet evidence:

• 828c4f0: cmux-ci compile-check --app passed. AgentPaneStateFileTests passed 9/9; UnsentDraftTests and ChatScrollRestoreDefaultPathTests also passed.
• Red/green: with the old shared static key (scratch commit 0f4c4a9), aReadOfAnotherFileFromInsideASaveWaitsForThatFilesSaves fails both of its expectations. At the head it passes.

[github-actions]

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.4100960879201 (a6724679c3)
• cmux NEXT nightly 1.0.0-nightly.4101142921301 (e1f81d2af5)