Docs
Short articles for a customer admin setting up and running a ProbeRun team. No ProbeRun internals are assumed.
| Article | For |
|---|---|
| Set up federated login (your IdP) | Connecting an OIDC identity provider, verified emails, group → role mapping, JIT |
| Set up SAML single sign-on | Your SAML IdP through the WorkOS Admin Portal, domain verification, connection state, who gets in |
| Invite and manage your team | Invitations, roles, removing members, what takes effect when |
| API keys and machine credentials | Creating keys, default expiry, rotation, what a key alone can and cannot do |
| The audit log | What is recorded, retention, paging and export bounds |
| Connect your agent over HTTP | RUNNER=http: the request ProbeRun sends, the reply shapes it accepts, a minimal server, testing on localhost |
| Using ProbeRun | Scenarios, Runs, Verdicts, failure clusters, calibration review, drift |
| Plan limits and fair use | Project allowance and its 402, the fair-use Run line that never blocks, reading your usage |
| Import a trace from Phoenix or Langfuse | Getting an exported agent run into probe import: the Phoenix span-search rewrap, the Langfuse collector tee |
Self-hosting and want a guided first run instead? The getting-started tutorial goes from a clean checkout to a real Verdict with a checkpoint at every step.