mjlab SKILL.md skillA

mjlab from nebius/nebius-physical-ai is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add nebius/nebius-physical-ai/mjlab@git:20261011.5082816

Pins git:20261011.5082816 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
nebius/nebius-physical-ai
Path
skills/tools/mjlab/SKILL.md
Versions recorded
6
Latest version
git:20261011.5082816, committed Oct 11, 2026
First version recorded
Jul 3, 2026
Size
6,532 bytes
Repository stars
106
License
Apache-2.0

Train, resume, evaluate, export, deploy or compose native MJLab policies through Workbench CLI, SDK, API and GPU workflows.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# MJLab

Read `docs/workbench/mjlab.md` for the supported MJLab 1.6.0 contract.
Behavior lives in `npa/src/npa/workbench/mjlab/`; CLI, SDK and HTTP clients share
that implementation. Keep simulator imports inside the isolated worker.

## Interfaces

- `npa workbench mjlab train`: upstream task defaults, optional iterations,
  environment count, learning rate, seed, native checkpoint resume and one-node
  multi-GPU torchrunx. The current process already owns its GPU allocation.
- `eval`: real complete episodes with measured returns, lengths and survival;
  `--video` publishes the first-episode MP4 and self-contained `rollout.html`
  with measured results and checkpoint/video hashes. `--device cpu` is supported
  for local evaluation.
  Native video defaults to 1920 × 1080 at the simulator step rate, encoded as
  H.264 CRF 18. `--video-width` / `--video-height` and the matching request fields
  accept even dimensions. The runtime checks decoded dimensions, timing and
  first-episode frame count; HTML reports the measured OpenGL renderer separately
  from CUDA simulation. Validate NVIDIA OpenGL on the actual rendering GPU.
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add nebius/nebius-physical-ai/mjlab@sha256:a2c8fc38031a1d87

The line at the top of this page pins the label git:20261011.5082816, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_3f7gbmpxyzqk6a3i.svg)](https://markdownregistry.com/a/art_3f7gbmpxyzqk6a3i)

3 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 6 versions of this file, between Jul 3, 2026 and Oct 11, 2026, about one every 20 days.

versioncommittedcommitsizeaudit
git:20261011.5082816 latest2026-10-11 5082816 6,532 BA view · diff
git:20260925.bf4788a2026-09-25 bf4788a 5,608 BA view · diff
git:20260906.525ee242026-09-06 525ee24 969 BA view · diff
git:20260802.5fa17712026-08-02 5fa1771 1,009 BA view · diff
git:20260724.349c7402026-07-24 349c740 995 BA view · diff
git:20260703.774344b2026-07-03 774344b 999 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (6532 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

nebius/nebius-physical-ai · 106 stars · license Apache-2.0 · pushed 2026-11-08 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_3f7gbmpxyzqk6a3i
GET https://markdownregistry.com/api/v1/resolve?ref=nebius/nebius-physical-ai/mjlab
GET https://markdownregistry.com/api/v1/blob/a2c8fc38031a1d87fd3860eaa35b5e0b4bc01910eddfa76bcb41121ee3f13c11

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from nebius/nebius-physical-ai

.cursorrules Cursor rule
nebius/nebius-physical-ai · .cursorrules
git:20260703.ce2e3a8 · audit A · 106 stars
AGENTS.md AGENTS.md file
nebius/nebius-physical-ai · AGENTS.md
git:20261106.20ba2be · audit A · 106 stars
CLAUDE.md CLAUDE.md file
nebius/nebius-physical-ai · CLAUDE.md
git:20261018.f0ea799 · audit A · 106 stars
DESIGN.md DESIGN.md file
nebius/nebius-physical-ai · DESIGN.md
git:20260906.525ee24 · audit A · 106 stars
access-approval SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/access-approval/SKILL.md · Prepare, verify, and resume operator-owned Hugging Face or NVIDIA NGC access for exact NPA dependencies, including…
git:20260912.c87fdb2 · audit A · 106 stars
agent-development SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/agent-development/SKILL.md · Use when building, enhancing, or testing the NPA chat agent backend — grounded-first routing, cost-aware Token Factory…
git:20261003.bd54c8c · audit A · 106 stars
agent-run-data-collection SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/agent-run-data-collection/SKILL.md · Use when collecting NPA agent work for fine-tuning or evaluation so every goal-level episode emits a sanitized…
git:20260906.2f687b4 · audit A · 106 stars
agent-visual-feedback SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/agent-visual-feedback/SKILL.md · Use when the NPA agent should describe or critique the current viewer (Rerun, video, image, or data) via the Describe…
git:20260906.4539e93 · audit A · 106 stars
architecture SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/architecture/SKILL.md · Use for Claude Code architectural review of the Nebius Physical AI platform, workbench layer, orchestrator choices, and…
git:20260906.2f687b4 · audit A · 106 stars
audit-container-docs SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/audit-container-docs/SKILL.md · Reconcile NPA container-image catalogs and related documentation whenever a workbench image or solution is added…
git:20260827.6716731 · audit A · 106 stars
build-and-push-image SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/build-and-push-image/SKILL.md · Use when building, tagging, validating, or publishing NPA workbench container images through immutable full-SHA public…
git:20260906.2f687b4 · audit A · 106 stars
contributions SKILL.md skill
nebius/nebius-physical-ai · skills/atomic/contributions/SKILL.md · Write or review NPA contributions, including workbench agent changes, using the repository's readability…
git:20260907.370e595 · audit A · 106 stars

Every file in nebius/nebius-physical-ai

Browse by kind, by grade A, or by owner.