Robot Skill Hub

Pinned robot code, environment requirements, preparation commands and execution evidence for coding agents.

Fetch the catalog

curl -fsSL https://hub.superfastrobotics.com/catalog/v1/index.json
curl -fsSL https://hub.superfastrobotics.com/agent-quickstart.md
curl -fsSL https://hub.superfastrobotics.com/llms-full.txt

• Catalog JSON: https://hub.superfastrobotics.com/catalog/v1/index.json
• Official source examples JSON: https://hub.superfastrobotics.com/catalog/v1/references.json
• Publication manifest: https://hub.superfastrobotics.com/catalog/v1/publication.json
• Agent quickstart: https://hub.superfastrobotics.com/agent-quickstart.md
• Discovery index: https://hub.superfastrobotics.com/llms.txt
• Full documentation: https://hub.superfastrobotics.com/llms-full.txt
• Record schema: https://hub.superfastrobotics.com/schemas/skill.schema.json
• Compatibility: https://hub.superfastrobotics.com/compatibility.md
• Contributions: https://hub.superfastrobotics.com/contributing.md

Find, inspect, prepare

Use the Python CLI for environment-aware search. It returns up to three candidates as JSON. Explicit environment mismatches are excluded; missing values remain unknown.

npx skills add Superfast-Robotics/robot-skill-hub --skill find-robot-skills

# Replace /PATH_TO_SKILL with the installed skill directory.
python /PATH_TO_SKILL/scripts/rsh.py --catalog https://hub.superfastrobotics.com/catalog/v1/index.json find "pick cube" --runtime maniskill --runtime-version 3.0.0b22 --robot panda --json
python /PATH_TO_SKILL/scripts/rsh.py --catalog https://hub.superfastrobotics.com/catalog/v1/index.json inspect maniskill-panda-pick-cube --json
python /PATH_TO_SKILL/scripts/rsh.py --catalog https://hub.superfastrobotics.com/catalog/v1/index.json prepare maniskill-panda-pick-cube --project ./experiment --plan --json

Inspect the selected record before preparation. --plan reports planned actions without downloads. Preparation and dependency installation are separate from simulation execution. Existing project locks remain pinned.

Hub preparation recipes (17)

• ManiSkill panda LiftPegUpright-v1 motion planner

  maniskill-panda-lift-peg-upright — Grasp a peg and lift it upright with the official Panda planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda_wristcam PegInsertionSide-v1 motion planner

  maniskill-panda-peg-insertion-side — Align a peg with a side-facing hole and insert it using Panda WristCam.

  maniskill 3.0.0b22, panda_wristcam, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill Panda PickCube Python recipe

  maniskill-panda-pick-cube — Grasp a cube and move it to a target with the official Panda motion planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: not_run; runtime: not_run; record version: contrib-a5dd78ad4e1d4e3d.

  Markdown | JSON

• ManiSkill panda PlaceSphere-v1 motion planner

  maniskill-panda-place-sphere — Pick up a sphere and place it into a bin with the official Panda planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda_wristcam PlugCharger-v1 motion planner

  maniskill-panda-plug-charger — Grasp a charger, align its pins, and plug it into a receptacle.

  maniskill 3.0.0b22, panda_wristcam, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda PullCubeTool-v1 motion planner

  maniskill-panda-pull-cube-tool — Use an L-shaped tool to pull a cube toward a goal.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda PullCube-v1 motion planner

  maniskill-panda-pull-cube — Pull a cube to its goal with the official Panda motion planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda PushCube-v1 motion planner

  maniskill-panda-push-cube — Push a cube to a goal region with the official Panda motion planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill Panda StackCube Python recipe

  maniskill-panda-stack-cube — Grasp one cube and stack it on another with the official Panda motion planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda StackPyramid-v1 motion planner

  maniskill-panda-stack-pyramid — Arrange three cubes into a pyramid using the official Panda planner.

  maniskill 3.0.0b22, panda, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda_stick DrawSVG-v1 motion planner

  maniskill-panda-stick-draw-svg — Trace the environment's default SVG outline with the Panda Stick planner.

  maniskill 3.0.0b22, panda_stick, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill panda_stick DrawTriangle-v1 motion planner

  maniskill-panda-stick-draw-triangle — Trace a triangle with the Panda Stick motion planner.

  maniskill 3.0.0b22, panda_stick, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill so100 PickCubeSO100-v1 motion planner

  maniskill-so100-pick-cube — Pick up a cube and move it to a goal with the official SO100 planner.

  maniskill 3.0.0b22, so100, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill xarm6_robotiq PickCube-v1 motion planner

  maniskill-xarm6-pick-cube — Grasp a cube and move it to a goal with the official xArm6 Robotiq solver.

  maniskill 3.0.0b22, xarm6_robotiq, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill xarm6_robotiq PlugCharger-v1 motion planner

  maniskill-xarm6-plug-charger — Insert a charger using the official xArm6 Robotiq motion planner.

  maniskill 3.0.0b22, xarm6_robotiq, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill xarm6_robotiq PushCube-v1 motion planner

  maniskill-xarm6-push-cube — Push a cube to a goal with the official xArm6 Robotiq solver.

  maniskill 3.0.0b22, xarm6_robotiq, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

• ManiSkill xarm6_robotiq StackCube-v1 motion planner

  maniskill-xarm6-stack-cube — Stack one cube on another with the official xArm6 Robotiq solver.

  maniskill 3.0.0b22, xarm6_robotiq, pd_joint_pos, state, linux

  Evidence: source_reviewed; runtime: not_run; record version: 1.0.0.

  Markdown | JSON

Official source examples (40)

These upstream examples are source references. They have no hub preparation adapter; follow the pinned upstream setup and invocation instructions. Fetch their complete metadata from https://hub.superfastrobotics.com/catalog/v1/references.json.

• MuJoCo minimal interactive simulator

  mujoco — Build the official minimal C++ simulator with GLFW camera interaction.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/sample/basic.cc

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo model format conversion

  mujoco — Convert MJCF, URDF, and binary model formats with the official compiler sample.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/sample/compile.cc

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo MJX tutorial

  mujoco — Use the official notebook to explore JAX-based accelerated MuJoCo simulation.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/mjx/tutorial.ipynb

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo offscreen trajectory recording

  mujoco — Render passive model dynamics offscreen and write raw frame data.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/sample/record.cc

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo full interactive simulator

  mujoco — Build and run the official simulator with model inspection, profiling, and sensor plots.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/simulate/main.cc

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo simulation benchmark sample

  mujoco — Time model rollouts and inspect contact, constraint, and CPU statistics.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/sample/testspeed.cc

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo Python tutorial

  mujoco — Work through official model loading, simulation, rendering, and control examples.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/python/tutorial.ipynb

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• MuJoCo Python interactive viewer

  mujoco — Launch the official Python viewer and inspect its passive and managed simulation interfaces.

  Source: https://github.com/google-deepmind/mujoco/blob/f1d45bd5422c74beddfb0d1deb590a02583d21de/python/mujoco/viewer.py

  License: Apache-2.0. Evidence: source_reviewed; runtime: not_run.

• PyBullet batch ray casting

  pybullet — Cast batches of rays around a simulated scene and visualize intersections.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/batchRayCast.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet collision filtering

  pybullet — Configure collision groups and masks between simulated bodies.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/collisionFilter.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet articulated constraints

  pybullet — Create and change constraints between robot links and other bodies.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/constraint.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet contact friction

  pybullet — Inspect contact behavior under different friction settings.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/contactFriction.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet force torque sensing

  pybullet — Enable joint force torque sensing and inspect the sensor readings.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/forcetorquesensor.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet simulation quickstart

  pybullet — Load a plane and R2D2 model, step simulation, and inspect the base pose.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/hello_pybullet.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet humanoid knee control

  pybullet — Explore position control on the knees of a simulated humanoid.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/humanoid_knee_position_control.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet end-effector orientation IK

  pybullet — Control KUKA end-effector position and orientation toward a fixed pose.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/ik_end_effector_orientation.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet inverse dynamics

  pybullet — Compute inverse dynamics and use torque control in a simulated articulated model.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/inverse_dynamics.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet KUKA inverse kinematics

  pybullet — Move a KUKA end effector along target positions using inverse kinematics and joint motors.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/inverse_kinematics.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet Jacobian computation

  pybullet — Compute a robot Jacobian and compare Cartesian velocity calculations.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/jacobian.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet KUKA grasp log playback

  pybullet — Replay a recorded KUKA grasp trajectory from the upstream state log.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/kuka_grasp_block_playback.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet KUKA and cube interaction

  pybullet — Run a KUKA arm along an IK trajectory in a cube scene and record state/contact logs.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/kuka_with_cube.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet PD control comparison

  pybullet — Compare proportional-derivative controllers on an articulated robot.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/pdControl.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet depth to point cloud

  pybullet — Convert rendered camera depth samples into a world-space point cloud.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/pointCloudFromCameraImage.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet quadruped simulation

  pybullet — Load and control a quadruped model using the upstream interactive example.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/quadruped.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet racecar control

  pybullet — Drive the simulated racecar using upstream motor and steering controls.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/racecar.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet differential racecar

  pybullet — Control a racecar model with differential drivetrain constraints.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/racecar_differential.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet interactive robot joint control

  pybullet — Inspect a robot's joints and drive them with GUI control parameters.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/robotcontrol.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• PyBullet save and restore state

  pybullet — Save and restore simulation states for repeatable experiments.

  Source: https://github.com/bulletphysics/bullet3/blob/2c204c49e56ed15ec5fcfa71d199ab6d6570b3f5/examples/pybullet/examples/saveRestoreState.py

  License: Zlib. Evidence: source_reviewed; runtime: not_run.

• robosuite collect and replay data

  robosuite — Collect state and action data, then replay it in the upstream demonstration workflow.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_collect_and_playback_data.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite composite robot

  robosuite — Instantiate a composed robot and exercise its action space in a manipulation task.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_composite_robot.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite controller demonstration

  robosuite — Compare controller behavior with prescribed action sequences.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_control.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite device teleoperation

  robosuite — Teleoperate a simulated robot with an upstream supported input device.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_device_control.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite domain randomization

  robosuite — Apply upstream domain randomization to visual and physical environment properties.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_domain_randomization.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite gripper interaction

  robosuite — Run an upstream gripper interaction test scene.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_gripper_interaction.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite gripper selection

  robosuite — Select robot and gripper models and simulate an upstream manipulation environment.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_gripper_selection.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite Gym wrapper

  robosuite — Use the upstream GymWrapper interface around a manipulation environment.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_gym_functionality.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite random action environment

  robosuite — Create an upstream robot manipulation environment and step random actions.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_random_action.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite segmentation observations

  robosuite — Render and inspect instance, class, or element segmentation observations.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_segmentation.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite sensor corruption

  robosuite — Add upstream sensor noise, delay, and corruption to observations.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_sensor_corruption.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

• robosuite video recording

  robosuite — Record a manipulation environment's camera frames to video.

  Source: https://github.com/ARISE-Initiative/robosuite/blob/a071383d53568ab798eb315c0e95357911be922d/robosuite/demos/demo_video_recording.py

  License: MIT. Evidence: source_reviewed; runtime: not_run.

Evidence

Each record keeps evidence.level, evidence.runtime_verification and compatibility status. not_run means no runtime execution evidence. source_reviewed does not mean the task succeeded.

Submit a useful fix

Allowed agents use https://api.superfastrobotics.com/v1/submissions with a scoped token and project sharing permission. The API creates an internal draft PR. Submission, review, merge and catalog publication are separate states. Read the contribution contract.