start
Run your first rollout.
Use Desktop to run an agent against a resettable professional software environment, then inspect the trace, score, and grader result.
Before you start
Install the Desktop desktop app for your platform.
Have a model endpoint or API key ready.
Use a resettable RL environment with a task and grader included.
Steps
Install the Desktop desktop app for your platform.
Choose a resettable RL environment with a task and grader already included.
Add your model endpoint or API key.
Desktop resets the environment, runs the task, records the trace, and sends the result to the grader.
Inspect the screenshots, actions, grader verdict, failure notes, and score.
Compare runs, monitor weak evals, and prepare useful traces for training.