fix: thread state and timeline fixes (upstream G4)

[rynfar]

Ports upstream group G4 (thread state and timeline) from T3 Code. The main behavior change: a thread whose agent finished but left a command running, such as a dev server, no longer reads as Waiting. It now shows as completed and unread. Only subagents and monitors still hold a thread at Waiting, because they will wake the agent again. The group also fixes three timeline and layout problems and shows the remote machine in the thread panel.

Sources

Upstream	Outcome	Notes
bb7997709d thread panel shows the remote machine a thread runs on (#14972)	Adopted (clean cherry-pick)	Uses Pylon's existing shouldShowEnvironmentIndicator. A lone non-primary environment gets a static label.
3a7058da50 new worktree threads no longer flicker while starting (#15019)	Adopted (import conflict resolved)	Pylon's ChatView.logic.ts imports ComposerSubmissionIntent only, so I kept that import and added derivePhase. Pylon's derivePhase is identical to upstream's.
db514607f4 sending on an older thread no longer jumps to the top (#15059)	Adopted (clean)	Runless folds are now decided per prompt, on both web and mobile.
a5b34b2537 a dev server left running no longer says the thread is waiting (#15114)	Adopted (test conflict resolved)	Pylon lacks upstream's subagent display-name tests, so I kept Pylon's test set and added upstream's new command-only presentation test.
f68e24fb41 a thread that left a shell running shows its unseen completion (#14910)	Adapted	Details below.
44b4965017 reduce the gap above the draft composer (#15196)	Adopted (clean)
a0bf1eee48 mobile: dev server running no longer shows the waiting bolt (#15194)	Adopted (clean)	FloatingWorkingStatus waiting → background with a waiting flag. It had no other Pylon consumers.

Pylon adaptations (f68e24fb41)

• ThreadNotificationCoordinator.tsx: skipped the upstream hunk and kept Pylon's version. Pylon's notifications go through projectAwarenessStates / reconcileAwarenessNotifications in packages/shared/src/agentAwareness.ts. That code reads the raw shell activityRunStatus ?? status and already uses backgroundWorkHoldsCompletion to treat a command-only roster as completed. Notification behavior and auto-settle are unchanged.
• Sidebar.logic.ts: minimal edits, kept well apart from G1's working-timer code.
  • Kept Pylon's status-model header comment.
  • Added one clause to the Working-shelf doc comment.
  • Took upstream's resolveThreadStatusPill change (backgroundWorkHoldsCompletion replaces the non-empty roster check).
  • Left isSidebarThreadWorking alone. It keeps the feat(web): beta Working sidebar section (upstream) #991 hasUnseenCompletion early return and the delegating/monitoring folding.
• Sidebar.logic.test.ts: adapted the upstream unseen completion with background work test. Upstream expects a monitor-waiting thread with an unseen completion in the Working shelf. Pylon's Working section (feat(web): beta Working sidebar section (upstream) #991) keeps every unseen completion in the inbox, so the test expects isSidebarThreadWorking === false for both rosters. The pill, top-status and recede expectations are unchanged.
• Client runtime (models.ts shellRuntime, threadExecution.ts deriveThreadRuntime) and mobile threadListV2.ts: adopted as upstream. A command-only roster no longer parks the runtime at idle.

Docs

• docs/user/status-indicators.md: explains that subagents, monitors and other background tasks keep a thread at Waiting, while commands left running (a dev server) do not. Such a thread shows as completed, unread until you open it.

Review follow-up

• Reworded the docs paragraph as above.
• Moved the mobile bolt-or-terminal icon choice into backgroundStatusSymbol in floating-working-status.ts and added a test for it. apps/mobile floating-working-status.test.ts passes (6 tests) and vp run -F @t3tools/mobile typecheck reports no errors.

Verification

• vp test run apps/web/src/components/ChatView.logic.test.ts apps/web/src/components/chat/MessagesTimeline.logic.test.ts apps/web/src/components/Sidebar.logic.test.ts packages/client-runtime/src/state/threadExecution.test.ts packages/client-runtime/src/state/entities.test.ts packages/shared/src/agentAwareness.test.ts packages/shared/src/orchestrationV2PendingBackgroundWork.test.ts: 7 files, 588 tests passed.
• apps/mobile: vp test run src/features/threads/threadListV2.test.ts src/lib/threadActivity.test.ts src/features/threads/floating-working-status.test.ts: 3 files, 185 tests passed.
• vp run -F @t3tools/web typecheck, vp run -F @t3tools/mobile typecheck and vp run -F @t3tools/client-runtime typecheck: each ran tsc --noEmit with no errors. Client-runtime printed only existing Effect suggestions in untouched tests.
• vp lint on the changed files: 7 warnings, the same 7 as on origin/pylon. None are in changed lines.
• vp fmt --check on the changed files passes, with one exception. When checked together with files from other packages, apps/mobile/src/features/threads/threadListV2.ts is flagged. It passes under vp fmt --check apps/mobile, and the unchanged base file is flagged the same way in that combined run.

Not verified

• No browser, simulator or device checks were run under the porting constraints, so there are no before/after captures. Not checked visually:
  • the composer gap (44b4965017)
  • the worktree-setup card and working-header layout (3a7058da50)
  • the thread panel environment row (bb7997709d)
  • the mobile terminal icon (a0bf1eee48)
• apps/desktop was not typechecked because it was not touched.

Part of upstream cycle #996.

🤖 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	4.9 KiB	−40 B (−0.8%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.7 KiB	3.7 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.1 KiB	−40 B (−3.3%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.4 KiB	20.4 KiB	−41 B (−0.2%)	29.3 KiB	✅
Codex	Live turn messages	2	1	−1 (−50.0%)	8	✅
Claude	Total thread wire	4.9 KiB	4.9 KiB	−41 B (−0.8%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.7 KiB	3.7 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	−41 B (−3.4%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	20.8 KiB	20.7 KiB	−41 B (−0.2%)	29.3 KiB	✅
Claude	Live turn messages	2	1	−1 (−50.0%)	8	✅

Baseline: 30ec43a · PR result: 98814d5 · 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: 106.1 KiB
• Claude decoded thread snapshot: 106.4 KiB

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