isaac-lab-3
Skill1 starsChanged 16 days agoApache-2.0View on GitHub
Use for Isaac Lab environments, manager terms, controllers, RL training, teleoperation, demonstrations, Mimic, imitation learning, resets, backend migration, and scaling on Antioch. Covers XYZW quaternions, ProxyArray, indexed writes, and PhysX/Newton selection; use Isaac Sim for plain scenes and scenario-design for recorded verdicts.
In this repository6 other files it gives its agents.
- agentic-simulationSkill
- antioch-platformSkill
- antioch-researchSkill
- isaac-sim-6Skill
- ros2Skill
- scenario-designSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
Use Lab's SimulationContext or environment loop, not antioch.world().
- Very specific: line 38
Check each field's type before accessing array methods.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Isaac Lab on Antioch
The pin is Isaac Lab 3.0.0-beta2 on Isaac Sim 6.0.1. Load Antioch platform for project/session setup and research for the selected runtime's signatures and source; beta APIs can change.
Startup and stepping
The scenario runner starts Lab through AppLauncher. A plain script or
notebook calls antioch.start_simulation() before simulator imports. Keep
isaaclab*, isaacsim, omni, pxr, and carb imports inside functions
or TYPE_CHECKING; define config classes in a post-startup factory so
discovery works without a simulator.
Use Lab's SimulationContext or environment loop, not antioch.world().
Configure SimulationCfg.dt, render interval, and environment decimation;
classic SimulationConfig.physics_dt/render_dt does not set Lab timing. Keep
one stepping owner.
Lab contracts
| Surface | Contract |
| Quaternions | XYZW; identity is (0, 0, 0, 1). |
| Array state | ProxyArray; access tensor methods through .torch or .warp. |
| Selected writes | write_*_to_sim_index or write_*_to_sim_mask; validate shape/device. |
| Cameras | Camera/CameraCfg; tiled names are deprecated aliases. |
| Physics | SimulationCfg.physics contains backend-specific configuration. |
| Launcher | Headless by default; visualizer options belong to the native launcher. |
Check each field's type before accessing array methods. Clone historical
samples and refresh views after a reset that replaces storage. In a manual
scene loop, write controls, step, then scene.update(sim.get_physics_dt());
a managed environment owns those updates.
Backends and training
Lab supports isaaclab_physx and isaaclab_newton. Select matching launch
and environment configuration before constructing the scene; PhysX
configuration comes from isaaclab_physx.physics.PhysxCfg. Newton-native
features do not imply support through every Lab asset, sensor, or solver,
and the managed Antioch path runs Kit; kit-less upstream examples are a
different launch path. Select images and dependencies through the platform's
environment guide.
Load environment authoring for state/reset contracts, a rollout probe, RL adapters, and demonstration workflows. Before scaling, verify observations, actions, frames, limits, termination conditions, and complete resets. Random actions test startup, not policy quality; finite loss or a checkpoint does not prove useful behavior.
Save seeds, configuration, asset/image pins, metrics, and checkpoints. GPU seeding does not guarantee identical runs. Use scenario design for recorded evaluations and agentic simulation for iterative experiments.
Load the part you need
| Task | Guide |
| Define manager terms, actions, observations, rewards, or episode resets | Environment structure and state |
| Check a shipped task before adding complexity | Small rollout probe |
| Register a task, choose an RL adapter, or load a checkpoint | Registration and RL |
| Record/replay demonstrations or generate Mimic data | Demonstrations and imitation |
| Scale environments while keeping a useful evaluation | Scaling and evaluation |
| Import a robot or validate its articulation | Conversion and articulations |
These guides adapt native workflow ownership to Antioch sessions. They do not replace Lab's native scene, manager, training, or reset APIs. Use the pinned upstream implementation through research when selecting an unfamiliar term or backend feature.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.