chore: merge upstream through ea7d46ac10

[andrewcai8]

Brings in the 7 upstream commits since #79: accessibility labels for controls and sidebar rows, usage contract tolerance for newer providers, no Cursor warning when no login is saved, client-version mismatch detection on the usage page, the offline-versus-update fix in the composer, and the Linux node-pty prebuild check for Windows artifacts.

Merge with a merge commit, not a squash. Squashing drops upstream's history and the next sync conflicts on every file again.

Conflicts

apps/server/src/usage/UsageService.ts. The fork reads every Cursor instance's login, and upstream now returns early when its one login is missing with no error. Kept the fork's multi-login loop and applied upstream's rule inside it. A login that is missing with no error adds neither an account source nor a setup error. A login blocked by an env token or the memory store still reports "needs a Cursor CLI login", as upstream does. Upstream's new tests pass unchanged. A new test covers two instances, one with an unreadable login and one with none, and expects only the unreadable login's error. Without the rule it gets two sources.

apps/web/src/components/ChatView.tsx. Upstream removed the folded "Reconnecting to X, finishing an update" line, which showed an offline server as an update in progress. Took that removal. Kept the fork's notice: the "no longer available / this workspace expired" state with its Connections action, and the Reconnect action. Reconnect stays visible while reconnecting, because a paused cloud workspace sits in that phase and only Reconnect wakes it (#33). The reconnectAction constant existed only so the folded line could share it, so it is inlined into its one remaining use.

apps/web/src/components/Sidebar.tsx. Kept the fork's pendingEnvironmentSend preview for draft rows, and added upstream's resolveSidebarRowAccessibility block after it, so the accessible label uses the same preview.

Verification

Targeted vp test run, all passing:

• apps/server src/usage/ has 8 files and 117 tests.
• apps/web covers ChatView.logic, Sidebar.logic, ui/switch, the usage page, and the usage query, with 13 files and 385 tests.
• packages/shared usageMerge.test.ts has 26 tests.
• packages/client-runtime usage and serverUsage have 11 tests.
• apps/mobile ConnectionFormField has 2 tests.
• scripts build-desktop-artifact has 75 tests.

No fork code reads the renamed staleEnvironments field.

🤖 Generated with Claude Code

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.5 KiB	+68 B (+0.5%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.1 KiB	7.1 KiB	−1 B (−0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.5 KiB	+69 B (+1.1%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	56.2 KiB	56.3 KiB	+88 B (+0.2%)	66.4 KiB	✅
Codex	Live turn messages	8	10	+2 (+25.0%)	21	✅
Claude	Total thread wire	13.5 KiB	13.5 KiB	+10 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.1 KiB	7.1 KiB	−5 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.5 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.0 KiB	57.0 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	9	9	0 (0.0%)	21	✅

Baseline: ca7e8c9 · PR result: b437bef · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 114.0 KiB
• Claude decoded thread snapshot: 114.7 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.