Installing a bundle creates a private copy you can edit. The published bundle stays unchanged.
109 bundles
Build environments where agents write SQL, execute it, and get scored on result correctness
Formally specify UI tasks with clear start states, goal states, and evaluation criteria
Trade-offs between pixel-level interaction and DOM-level interaction for UI agents
Build instrumented browser environments with action logging and state capture
Build environments where agents navigate and modify entire repositories, not just single files
Use test pass rates as automatic reward signals for code generation
Score code changes on correctness, style, security, and performance
Build environments for training code completion models (à la Cursor's online RL)
Maintain and curate experience replay buffers for continual RL training
Detect when the production task distribution has drifted from the training distribution
Prevent RL training from destroying previously learned capabilities
Set up learning loops where production experience feeds back into training