| name | monitoring |
| description | Monitor Hawk job status, view logs, and diagnose issues. Use when the user wants to check job progress, view error logs, debug a failing job, or generate a monitoring report for a Hawk evaluation run. |
Hawk Job Monitoring
Monitor running or completed Hawk jobs using the hawk logs command or hawk status subcommands.
Job ID
The JOB_ID parameter is the eval_set_id or scan_run_id from when the job was submitted.
JOB_ID is optional. If omitted, uses the last eval set ID that was used or received.
Available Commands
1. View Logs (Shorthand)
The hawk logs command shows logs:
hawk logs
hawk logs <JOB_ID>
hawk logs -n 50
Options:
-n, --lines N - Number of lines to show (default: 100)
Note: Do NOT use the -f/--follow flag - it blocks indefinitely and is intended for interactive terminal use only.
2. View status
Generate a full monitoring report with logs, metrics, and other details:
hawk status
hawk status <JOB_ID>
hawk status <JOB_ID> > report.json
Options:
--hours {hours} - Fetch logs from last N hours (default: 24 hours)
3. Watch live status
hawk watch shows live per-task / per-sample progress: phase, retries, limits, scores, and pod/scheduling trouble. Best for "is it stuck, and why":
hawk watch
hawk watch <JOB_ID> --no-follow
hawk watch <JOB_ID> --json
Note: Plain hawk watch streams until the run finishes — use --no-follow or --json for a single snapshot in non-interactive use.
Common Workflows
Check job progress
hawk logs
Triage a stuck or slow job
hawk watch <JOB_ID> --no-follow
See the debug-stuck-eval skill for interpreting the output.
Generate full report for analysis
hawk status > report.status