# OORB RunFP documentation

Practical guides to robot run evidence, comparisons, release checks, and coding agent integration. Access to RunFP is arranged through OORB onboarding.

Canonical: https://oorb.io/docs/

## Start here

- [Get started with RunFP](https://oorb.io/docs/getting-started/): Initialize a local workspace, import your own MCAP recordings, and set a comparison baseline with OORB RunFP.

## Work with runs

- [Record robot run provenance](https://oorb.io/docs/robot-run-provenance/): Keep telemetry and captured software, hardware, and model evidence together in comparable robot episodes.
- [Compare robot runs](https://oorb.io/docs/compare-robot-runs/): Compare two stored episodes, inspect recorded changes, and trace outcomes alongside workspace commits.

## Evaluate releases

- [Measure robot task success rates](https://oorb.io/docs/robot-success-rate/): Choose recorded outcomes or comparison criteria, report a Wilson confidence interval, and interpret coverage limits.
- [Check robot regressions in CI](https://oorb.io/docs/robot-regression-ci/): Compare a new MCAP recording with a named baseline and gate releases using explicit comparison or outcome criteria.

## Integrations

- [Track robot model provenance](https://oorb.io/docs/model-provenance/): Record local model identity or a Weights & Biases artifact and inspect the model evidence associated with runs.
- [Prepare observed Real2Sim replay](https://oorb.io/docs/real2sim-replay/): Package an episode and robot description for replay readiness checks in an Isaac Sim and ROS environment.
- [Connect RunFP to coding agents](https://oorb.io/agents/): Connect Codex or Claude Code to OORB's local MCP server to inspect projects, saved robot scans, and run comparisons.