ProbeRun

Docs

Using ProbeRun

ProbeRun is scenario-based regression testing for AI agents. You describe how your agent should behave, ProbeRun runs the agent and grades the result, and then it tells you what changed between one run and the next. The words below are the product’s own vocabulary; the dashboard and API use them consistently.

  1. Light and dark
  2. Finding your way
  3. Running your own server
  4. Your first Project
  5. Scenarios
  6. Runs, Trajectories and Verdicts
  7. Reading a Project page
  8. Calibration review
  9. Comparing runs and watching for drift
  10. Policies and Violations
  11. Plan limits and fair use
  12. How long Runs are kept
  13. SAML single sign-on
  14. API keys
  15. Large lists and paging