Run Statuses
- Queued: Waiting for worker execution.
- Running: Worker is executing the agent.
- Completed: Successful execution; full result available.
- Failed: Terminated due to error; error message displayed on detail page.
- Timed out: Exceeded configured per-run limit.
Run History
- Access: Navigation: Scheduled Jobs -> View Runs (displays 50 most recent runs, auto-polling).
- Trigger manual run: Click Run Now.
- View details: Click any run row.
Stopping a Run
- Applicable statuses: Queued or Running only.
- Method 1 (History List): Click red stop icon next to the status badge.
- Method 2 (Detail Page): Click red Stop Job button in the header.
- Resulting status: Failed (Error:
Job killed by user). Auto-marked as read. Captured telemetry is preserved.
Run Detail Page Sections
- Prompt: The active prompt configuration at run time.
- Thinking: Numbered step-by-step execution logs of the agent.
- Tool Calls: Sequential log of tools executed:
- Tool name (
web-browsing, web-scraping, custom skills).
- Call timestamp.
- Input arguments (JSON format).
- Output payloads (via Show result toggle).
- Files: Generated artifacts (exports, charts, docs) available for download.
- Warning: Files generated in single-user mode become inaccessible via the UI if the instance is migrated to multi-user mode.
- Response: Final markdown output from the agent.
- Metrics: Token usage (prompt and completion tokens).
Continuing a Run in a Workspace Thread
- Condition: Run must be in Completed status.
- Action: Click Continue in Thread.
- Process:
- System creates/opens the Scheduled Jobs workspace.
- System instantiates a new thread containing the job prompt and the final response.
- User is redirected to the thread to resume interactive chat using active workspace tools/settings.
Read/Unread Status
- Opening a run’s detail page clears its unread indicator. Stopped runs are marked read automatically.