MJLab

SkillDev tools

Use when working on MJLab locomotion evaluation, SONIC checkpoint scoring, SkyPilot MJLab YAMLs, or Workbench MJLab CLI behavior.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the MJLab skill

What this skill tells your AI

The instructions your AI receives, as published by nebius/nebius-physical-ai in skills/tools/mjlab/SKILL.md and read by ahel’s review.

MJLab is the locomotion evaluation stage for SONIC Workbench workflows.

Interfaces

CLI:

npa workbench mjlab eval
npa workbench mjlab workflow
npa workbench mjlab status
npa workbench mjlab list

SkyPilot YAML:

  • workflows/testing/mjlab-eval.yaml
  • workflows/testing/sonic-locomotion-finetuning.yaml

Routing And Data Flow

Route MJLab evaluation to H100 for the checked-in workflow templates.

Inputs and outputs use S3 paths:

  • --input-path: retargeted motion or rollout artifacts.
  • --checkpoint: SONIC checkpoint artifact.
  • --output-path: MJLab evaluation output prefix.

The result artifact is mjlab_eval.json.

Workflow Constraint

Keep orchestration logic in SkyPilot YAML. Do not add a Python runner script for the SONIC locomotion fine-tuning path.

Signals

GitHub stars
28
Forks
15
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
mjlab
Source
github.com/nebius/nebius-physical-ai