fix(web): a command you stop stays in the timeline

[juliusmarminge]

When you stop a turn while a command is running, the command disappears from the timeline once the run settles. You see "Running sleep" while it runs, then after Stop only the "Run interrupted" marker is left, with no trace of the command that was cut short.

Seen live on Codex 0.156.1 (gpt-6-luna). The prompt was sleep 90 && echo slept, stopped mid-command. The projection keeps the command_execution turn item (status: "interrupted", input: "/bin/bash -lc 'sleep 90 && echo slept'", no output), but the DOM has no row for it:

Why

interrupted and cancelled items map to the work-log status stopped. workEntryIsVisibleInGroup hides tool rows that are "neutral", meaning neither success nor failure, which is meant for empty or incomplete rows. workEntryIndicatesToolSuccess explicitly excludes stopped, so every stopped tool counted as neutral and was filtered out of settled rows. The labels already have a stopped outcome ("Stopped vp", "Stopped clicking in the preview browser"), and mobile's threadFeedActivityIsVisible already keeps lifecycleStatus === "stopped" rows. Only web dropped them.

Fix

workEntryIsVisibleInGroup keeps entries whose lifecycle status is stopped. Nothing else changes: success and failure markers, folds, and in-progress rows behave as before.

Verification

• New case in MessagesTimeline.logic.test.ts ("hides the interruption request while keeping intervening work and the result"): it builds a stopped command followed by the interrupt result for an interrupted latest run. It fails on the base (expected [ 'interrupt-result' ] to deeply equal [ 'stopped-command-entry', … ]) and passes with the fix.
• vp test run src/components/chat/MessagesTimeline.logic.test.ts src/session-logic.test.ts: 177 passed.
• vp test run src/components/chat/MessagesTimeline.test.tsx src/components/chat/timelineMinimapItems.test.ts src/session-logic.runtime-diagnostics.test.ts: 81 passed.
• vp exec tsc --noEmit -p . in apps/web: clean. vp lint and vp fmt on the two touched files: clean.
• Not captured: an "after" screenshot. The hunting dev server has to keep running the base code, and cups was short on memory for a second one. The replayed test covers the row.
• Not run: the full web suite.

Model: Claude Opus 5.5 (Claude Code)

🤖 Generated with Claude Code

---

[macroscopeapp]

Approvability

Verdict: Approved at a5fa963

Macroscope's review found this PR approvable — This is a narrowly scoped web timeline bug fix that preserves stopped command rows without changing command execution or other lifecycle behavior. A focused regression test covers the interrupted-command case, and no product-default or static-analysis configuration is modified.

^{You can add or adjust custom eligibility rules. Learn more.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	4.9 KiB	—	6.8 KiB	✅
Codex	Thread snapshot wire	—	3.7 KiB	—	4.9 KiB	✅
Codex	Live turn WebSocket wire	—	1.2 KiB	—	2.0 KiB	✅
Codex	Live turn WebSocket decoded	—	20.4 KiB	—	29.3 KiB	✅
Codex	Live turn messages	—	2	—	8	✅
Claude	Total thread wire	—	4.9 KiB	—	6.8 KiB	✅
Claude	Thread snapshot wire	—	3.7 KiB	—	4.9 KiB	✅
Claude	Live turn WebSocket wire	—	1.2 KiB	—	2.0 KiB	✅
Claude	Live turn WebSocket decoded	—	20.8 KiB	—	29.3 KiB	✅
Claude	Live turn messages	—	2	—	8	✅

Baseline: unavailable · PR result: a5fa963 · 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.