executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
mimic-kitUse MimicKit repo-specific guidance for motion imitation, physics-simulator RL training, motion conversion, and SMP prior workflows.
Independent GitHub snapshot · Automated classification · Original documentation governs
Use this skill when a task is about MimicKit, a checkout-oriented motion imitation and physics-simulator control framework. It covers the repository's runner, simulator backend choices, motion data tools, DeepMimic/AWR/LCP recipes, AMP/ADD/ASE recipes, and Score-Matching Motion Prior workflows.
Read references/repo-provenance.md before deciding whether this skill is current for a checkout. Read references/troubleshooting.md when the request starts with an install, backend, missing asset, or config-path failure.
view_motion, DoF diagnostics, or plotting logs, load motion-tools.smp_prior_cfg, smp_prior_model, or SMP task policies, load smp.pyproject.toml, setup.py, or setup.cfg in this checkout. Treat it as a checkout-oriented project, not as a pip-installable distribution.mimickit/ directory to be importable. The bundled runner wrapper in sub-skills/runner-and-backends/scripts/run_mimickit.py prepares that import layout.requirements.txt, but simulator backends are external: Isaac Gym, Isaac Lab/Isaac Sim, or Newton/Warp must be installed separately according to the chosen engine.For a target checkout, start with the non-simulator checker:
python sub-skills/runner-and-backends/scripts/check_mimickit_layout.py \
--repo-root <mimickit-checkout>
To preview a runner command without launching a simulator:
python sub-skills/runner-and-backends/scripts/run_mimickit.py \
--repo-root <mimickit-checkout> \
--dry-run \
-- --arg_file args/deepmimic_humanoid_ppo_args.txt --visualize false
For motion tools, prefer the bundled helpers in sub-skills/motion-tools/scripts/ instead of source utility paths. For SMP prior/policy checks, use sub-skills/smp/scripts/train_smp_prior.py --dry-run-config and sub-skills/smp/scripts/check_smp_config.py before launching training.
This generated skill was verified for source syntax/import routing, PyTorch CUDA availability, converter parser help, tiny GMR/SMPL conversion fixtures, plotting helper output, and SMP dry-run/config checks. It was not verified by running full simulator-native training/testing because the external simulator packages and downloaded data/model/object assets were unavailable in the production environment.
Do not auto-import or claim full backend validation from this skill alone. If a user needs full native validation, install the selected simulator backend, restore the required data/model/object assets, and run a focused native case after checking the relevant sub-skill.