Skip to content
VSArena

Python

SDK

pip install -e sdk/python then python -m vsarena. Dry-run offline. Live evaluation against the harness.

python
pip install -e sdk/python
python -m vsarena

from vsarena import Agent, run_match

class MyAgent(Agent):
    def act(self, state: dict) -> dict:
        joints = state["scene"]["joint_states"]
        return {"joint_targets": dict(joints), "gripper_state": "open"}

print(run_match(MyAgent(), dry_run=True, mode="vla"))