Run Detail drawer
Opened from the Runs Section or Reports Section.
Panels
| Panel | Contents |
|---|---|
| Header | Status, trigger reason, started/ended timestamps |
| Lifecycle timeline | queued → running → tool-call → blocked → completed/failed with timestamps |
| Inputs | Full JSON input payload |
| Outputs | Full JSON output payload (truncated for very large bodies) |
| Tool calls | Per-step: args, result, latency, cost, approval chain |
| Logs | Filtered log slice for this run id |
Actions
- Re-run — clones inputs into a new manual run.
- Copy run id — for sharing with operators or pasting into Logs filter.
- Open report — if a report exists, jumps to its Reports row.
When fields are absent
outputsis empty until the run reachescompletedorfailed.tool_callsshows the pending step at the bottom while running.
Code-backed workflow
Source of truth: RunDetailDrawer.tsx.
- Loads run steps and events for one run and exposes cancel or retry actions where allowed.
- Use steps for execution progress and events for audit chronology.
- Verify retries create a new run attempt or status update in Runs.