Installing a bundle creates a private copy you can edit. The published bundle stays unchanged.
109 bundles
Score intermediate reasoning steps, not just final outcomes
Combine multiple reward signals (correctness, efficiency, style, safety) into a single scalar
Use a language model to score agent outputs against specifications or rubrics
Translate qualitative rubrics into multi-dimensional scoring functions with partial credit
Design pass/fail reward signals (code compiles, test passes, form submitted correctly)
Build high-quality static datasets from historical trajectories for offline RL or behavior cloning
Convert heterogeneous log formats into a unified trajectory schema (state, action, reward, metadata)
Strip PII, credentials, and sensitive business data from trajectories while preserving RL-relevant structure
Score and filter trajectories by quality, remove corrupted/incomplete episodes
Log agent rollouts with full state-action-reward-next_state tuples, tool calls, and timing
Instrument production tools to log human expert actions, states, and outcomes as RL-ready trajectories
Break complex enterprise workflows into subtask chains with intermediate checkpoints