feat(web): surface orchestration agents in the timeline

[shivamhwp]

Orchestrator V2 already projects native subagents, and the existing Agents panel already owns their live detail. The remaining problem is presentation: chat repeats one lifecycle row per agent, duplicating the roster and stretching the turn.

This PR keeps the existing orchestration semantics and:

• collapses only local subagents from the same run into one compact timeline action;
• derives the visible count and aggregate state from existing projected item statuses;
• keeps partial failures visible and destructive while sibling agents are still working;
• opens the existing Agents panel for names, status, duration, output, and hierarchy;
• preserves inherited fork projections as lifecycle rows so they keep their source-thread context;
• keeps superseded and replacement attempts in separate summary rosters;
• isolates changing agent summaries from unrelated virtualized timeline rows.

There are no server, contract, persistence, provider-adapter, recovery, client-runtime, or mobile behavior changes.

Screenshots

Before · individual agent rows	After · summaries and Agents panel

Narrow timeline	Narrow Agents panel

Verification

• 169 focused timeline tests pass across two files.
• Web typecheck passes.
• The changed timeline files pass formatting, lint, and whitespace checks.
• Render coverage confirms local agents collapse to one summary row while inherited rows and separate attempts remain independent.
• The summary exposes working, idle, failed, stopped, and completed states and opens the existing Agents panel.

Built and tested with GPT-5.6-Sol in the Codex harness through T3 Code.

---

Note

Low Risk
UI-only timeline presentation and row derivation; no auth, API, or persistence changes, with a safe fallback when the Agents opener is omitted.

Overview
Local subagent events in the chat timeline are collapsed into one summary row per run (or per attempt / orphan item) instead of repeating a lifecycle card for every agent, while inherited fork projections still render as normal event rows.

deriveMessagesTimelineRows gains optional summarizeSubagents and builds agent-spawn rows with aggregate working, idle, failed, and stopped counts; non-representative entries in a group are dropped from the list. Summarization turns on only when MessagesTimeline receives onOpenAgents—ChatView wires that to the existing Agents panel opener.

The new AgentSpawnCtaRow is a single clickable control (“Kicked off N agents…”, failure styling when needed, Open Agents / View) so roster detail stays in the Agents panel. Without onOpenAgents, behavior stays the old per-event rows. Stable row comparison and tests cover grouping, idle vs completed, superseded attempts, and cache refresh when agent status changes.

^{Reviewed by Cursor Bugbot for commit a61796e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add grouped agent-spawn summary rows to MessagesTimeline timeline

• ChatView passes its existing Agents-surface opener to MessagesTimeline via a new optional callback prop, enabling subagent summarization only when the callback is present
• deriveAgentSpawnSummaries in MessagesTimeline.logic.ts aggregates local subagent events into per-run, per-attempt, or per-item groups, suppressing later entries in each group and exposing working, idle, failed, and stopped counts
• The new AgentSpawnCtaRow component renders each summary as a clickable control showing total and status counts; active groups offer "Open Agents" and settled groups offer "View", with destructive styling for failures
• Inherited entries, non-subagent events, and entries without a summary remain ordinary event rows
• Behavioral Change: MessagesTimeline now displays grouped summary rows instead of individual local subagent events when the Agents opener is provided; without it, rendering is unchanged. The new agent-spawn variant in MessagesTimelineRow must be handled by any consumer of that union

^{Macroscope summarized 62d1993.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Team

Run ID: 606ccf48-8344-45dd-ab00-72d4e58d33f9

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch subagent-obs/05-thread-visibility

Warning

Your free Security trial is over. An organization admin can upgrade to Advanced for continuous pull request security review or dismiss this notice.

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

One finding on the new Agents CTA row in MessagesTimeline.tsx: its trailing metadata group can't shrink, so a provider-supplied workflow phase title overflows the row on narrow chat columns. Details inline.

^{Posted via Macroscope — UI Consistency}

[cursor]

Stale Bugbot comment from a previous run.

[macroscopeapp]

Macroscope has since reviewed this pull request. An earlier review was skipped by a cost limit; a review has now completed, so that notice no longer applies.

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 6f53d1a. Configure here.}

[cursor]

Bugbot is paused — on-demand spend limit reached

Bugbot uses usage-based billing for this team and has hit its on-demand spend limit.

A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue.