native/source/task.skill.json

{
  "vdia": "0",
  "skill": "place_cup",
  "version": "0.3.0",
  "task": "Set a cup on the mark.",
  "rail": "physical",
  "demos": 1,
  "format": "lerobot/v3",
  "codebase_version": "v3.0",
  "robot_type": "dvidia.wrist_room",
  "cameras": ["ego", "room"],
  "residual": {"claimer": 0.5, "contributors": 0.5, "book": "not-live"}
}