chore: merge upstream pingdotgg/t3code through 42da166d13

[andrewcai8]

Main was 39 upstream commits behind after #204. This merges upstream 7ac952c630..42da166d13 as a real merge commit. Merge with Create a merge commit, never squash.

Notable upstream changes

• Subagents. Steering no longer ends running subagents (fix(server): steering no longer ends running subagents pingdotgg/t3code#19210). A delegated task's question wakes its parent and ends the parent's blocking wait (fix(server): a delegated task's question now wakes its parent agent pingdotgg/t3code#19521, fix(server): a delegated task's question ends the parent's blocking wait pingdotgg/t3code#19538). A thread comes back when its subagent needs approval (fix(clients): a thread comes back when its subagent needs approval pingdotgg/t3code#19630).
• Server. Cold usage scans no longer stall the server (fix(usage): cold usage scans no longer stall the server pingdotgg/t3code#19446). Commands on busy threads stop decoding finished tool nodes (perf(server): commands on busy threads stop decoding thousands of finished tool nodes pingdotgg/t3code#19675). Restarts no longer stall on old subagent results (fix(server): restarts no longer stall on old subagent result deliveries pingdotgg/t3code#19631).
• Codex. "Always allow this session" stops asking for every command (fix(server): "Always allow this session" stops Codex asking for every command pingdotgg/t3code#19635).
• Text generation no longer falls back to the flagship model (fix(web): text generation no longer falls back to the expensive flagship model pingdotgg/t3code#19722).
• New behavior. PR branches can be deleted after merge or close (feat(pr): delete branches after merge or close pingdotgg/t3code#19257). Settled thread worktrees get cleaned up (feat(storage): clean up settled thread worktrees pingdotgg/t3code#15208). Auto-settle can keep threads with a PR (feat(settings): inactivity auto-settle can keep threads with a PR pingdotgg/t3code#14642).

Conflicts

• settings.ts and SettingsPanels.tsx. Both sides added a field next to each other. Both stay.
• threadSettled.ts. Keeps the fork's re-export of threadSettlement.ts and upstream's new import for subagent requests.
• Sidebar.logic.ts. Keeps the fork's workspace-missing check and takes upstream's subagent approval check.
• ChatView.tsx and Sidebar.tsx. Upstream turned callbacks into plain functions wrapped in useStableCallback (fix(web): the chat view and composer no longer keep old renders alive pingdotgg/t3code#19869, fix(web): the sidebar no longer keeps old renders alive pingdotgg/t3code#19852). The fork's cloud additions move into the new functions: cloud.onEnvironmentPicked, cross-environment related-thread routing, cloudRunOn, footerCard and startsCloudEnvironment. The environment picker still shows for cloud run targets. The fork's dependency arrays are gone.
• Sidebar.tsx status. The fork keeps its SIDEBAR_TOP_STATUS table. Upstream's new Waiting icon (feat(web,mobile): waiting status shows a dot-dashed icon and a timer pingdotgg/t3code#19617) moves into it.
• ProviderTurnStartService.test.ts. Takes upstream's resolveRuntimePolicy mock and fixture fields. The fork's delivered-turn tests keep a real temp directory as cwd, and the fixture keeps the fork's lineage.

Fork code adapted to upstream

• Fork fakes of the Claude query session gain backgroundTasks, which upstream added to ClaudeAgentSdkQuerySession.
• Upstream's new StorageSettings test fixture includes the fork's cloudMachinesAfterDays.

Verification

• Every line the fork added in a file both sides changed is still present, except the replaced dependency arrays, test stub and Waiting entry listed above.
• Server: ProviderTurnStartService, ForkMigrations and SessionStore, 59 tests pass. Web: Sidebar, ChatView and settings, 70 files, 1034 tests pass. client-runtime state: 65 files, 886 tests pass.
• vp i leaves pnpm-lock.yaml unchanged. CI is green, including typecheck. One earlier CI run timed out in remotePreparation.test.ts's native addon build; the file is unchanged by this merge, and it passed on the rerun.

Claude Opus 5.5, Claude Code

🤖 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	4.9 KiB	5.0 KiB	+23 B (+0.5%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	+23 B (+1.9%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.8 KiB	20.9 KiB	+41 B (+0.2%)	29.3 KiB	✅
Codex	Live turn messages	1	2	+1 (+100.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	0 B (0.0%)	29.3 KiB	✅
Claude	Live turn messages	2	2	0 (0.0%)	8	✅

Baseline: f642d2a · PR result: c6b5d9a · 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: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

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