Skip to main content
The Critiqor CLI is the primary interface for attaching runtime observation to OpenClaw agents, generating diagnosis reports, and reviewing reliability results. Commands are grouped around a simple lifecycle: start observing → interact with your agent → finalize → review. Each command maps to a specific phase of that workflow, and all run artifacts are stored under a local runs/ directory by default.

​
Commands

CommandPurpose
critiqor monitor openclawLaunch OpenClaw with runtime observation active
critiqor finalizeStop observation, generate diagnosis, open dashboard
critiqor dashboard [run_id]Open the local dashboard for the latest or a specific run
critiqor runsList all completed runs with summaries
critiqor checkRun a deployment policy gate (for CI/CD)

​
Basic workflow

critiqor monitor openclaw   # start observing
# ... interact with your agent ...
critiqor finalize           # generate diagnosis + open dashboard
Once critiqor finalize completes, the local dashboard opens automatically in your browser. You can re-open it at any time with critiqor dashboard, and browse all past runs with critiqor runs.
If a monitoring session is already active, running critiqor monitor openclaw again will exit immediately and prompt you to run critiqor finalize first. Only one session can be active at a time.

​
Command reference pages

  • monitor openclaw — Start observing an OpenClaw session
  • finalize — Generate a diagnosis and open the dashboard
  • dashboard — Re-open or switch between run dashboards
  • runs — List all completed evaluation runs
  • critiqor check — Run a deployment policy gate (use critiqor check --help for options)