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"))