Agents train the policy. A person approves it onto the robot.
2robots on the hub
Add a robot from a zip
Drop an MuJoCo's robot description file (XML); the hub simulates and renders from it. or ROS's robot description file (XML); the hub converts it to MJCF when you add a robot. with its meshes; the hub checks the card and its The robot's own declaration of what it is to a policy: joint order, base, default pose, gains, what an action means, the control rate. before anything is published.
Train in the cloud or on your own machine
A An API key made with one scope — read, write or train — so whatever holds it acts as you only that far. submits the One training job: a template, a robot, a few settings and a budget; it ends in a model or a failure, with its log and checkpoints kept.; it trains on Modal where a deployment names an app, or on a A small program you start on your own computer; it picks up your queued jobs and runs them there. you start on your own GPU.
Gate it in simulation
Every A snapshot of the policy saved partway through training, so an earlier point can be tested or deployed. scores PASS or FAIL on a The set of test cells a rollout runs: smoke is a few quick ones, full is the whole table. before it is anything but a file.
Deploy with a hold, stop with a click
A signed-in person holds to approve one The trained program that turns what the robot senses into the motor commands it sends — what a run produces and a robot runs. onto one robot for a window, and one click stops it. No key can do either.
Keys for your agents
An API key made with one scope — read, write or train — so whatever holds it acts as you only that far. for the CLI and the The protocol coding agents (Claude Code, Cursor) use to call tools; the hub ships a server for it. server — read, write, train — minted under your name.