evaluating-cosmos-policy

evaluating-cosmos-policy is a skill for Claude Code from Orchestra-Research/AI-Research-SKILLs. It costs 51 tokens per session (3,729 once invoked), scanned A, original, MIT.

A workflow for evaluating NVIDIA Cosmos Policy, a robot-control model, in simulated LIBERO and RoboCasa environments. It includes headless GPU runs, which perform evaluations without opening a display.

In plain words
What is it for?
Use it to set up simulation evaluations, run GPU-based tests without a screen, and profile robot-policy latency.
Why use it?
It provides a repeatable way to test robot manipulation policies and measure their inference speed on local or cluster GPUs.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the multimodal plugin — 10 skills shipped together

Good fit Use it to set up simulation evaluations, run GPU-based tests without a screen, and profile robot-policy latency.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/orchestra-research/ai-research-skills/cosmos-policy
About the project

AI Research Skills Library is a collection of reusable instructions that guide AI agents through research and machine-learning engineering tasks, from finding ideas and writing papers to training, evaluation, and deployment. It is for configuring agents such as Claude Code, Codex, and Gemini to perform research workflows.

Orchestra-Research/AI-Research-SKILLs · 12,567 stars · on GitHub · orchestra-research.com

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add Orchestra-Research/AI-Research-SKILLs --skill cosmos-policy
Clone the repo
git clone --depth 1 https://github.com/Orchestra-Research/AI-Research-SKILLs

Made for: Claude Code.

Or install multimodal, the plugin that ships this one along with the rest of its 10 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for evaluating-cosmos-policy

README.md
[![agentmods](https://agentmods.dev/badge/skills/orchestra-research/ai-research-skills/cosmos-policy/github.svg)](https://agentmods.dev/skills/orchestra-research/ai-research-skills/cosmos-policy)
Your own site
<a href="https://agentmods.dev/skills/orchestra-research/ai-research-skills/cosmos-policy"><img src="https://agentmods.dev/badge/skills/orchestra-research/ai-research-skills/cosmos-policy/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for evaluating-cosmos-policy

Your own site · 80×15
<a href="https://agentmods.dev/skills/orchestra-research/ai-research-skills/cosmos-policy"><img src="https://agentmods.dev/badge/skills/orchestra-research/ai-research-skills/cosmos-policy.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 51 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,729 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 2 findings, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium Rogue Agent · line 353
    Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
    Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
  • low Excessive Agency · line 272
    Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.
    Fix: Limit the skill's scope to its documented purpose. Remove instructions that enable the agent to perform actions outside its stated functionality.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00051 $0.03729
Opus 5 $0.00026 $0.01865
Sonnet 5 $0.00010 $0.00746
Haiku 4.5 $0.00005 $0.00373

Measured 9d ago against content hash 31d6b4701fae, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

evaluating-cosmos-policy scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

18-multimodal/cosmos-policy/SKILL.md · 403 lines

How it starts

The opening of the file, as written. The whole thing — 403 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cosmos Policy Evaluation

Evaluation workflows for NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments from the public cosmos-policy repository. Covers blank-machine setup, headless GPU evaluation, and inference profiling.

Quick start

Run a minimal LIBERO evaluation using the official public eval module:

uv run --extra cu128 --group libero --python 3.10 \
  python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
    --config cosmos_predict2_2b_480p_libero__inference_only \
    --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 16 \
    --num_open_loop_steps 16 \
    --task_suite_name libero_10 \
    --num_trials_per_task 1 \
    --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --ar_future_prediction False \
    --ar_value_prediction False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

Core concepts

What Cosmos Policy is: NVIDIA Cosmos Policy is a vision-language-action (VLA) model that uses Cosmos Tokenizer to encode visual observations into discrete tokens, then predicts robot actions conditioned on language instructions and visual context.

Key architecture choices:

Component Design
Visual encoder Cosmos Tokenizer (discrete tokens)
Language conditioning Cross-attention to language embeddings
Action prediction Autoregressive action token generation

Read the full file on GitHub · 403 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 9d ago First seen · 403 lines · 51 tokens per session scan A 31d6b4701fae

Subscribe to this mod's changes

evaluating-cosmos-policy is a skill published in the GitHub repository Orchestra-Research/AI-Research-SKILLs (12,567 stars, last pushed 2mo ago), licensed MIT. It adds 51 tokens to every session and 3,729 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Related

Other skills, from other repositories

research-cosmos-policy

Evaluates NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments. Use when setting up cosmos-policy for robot manipulation evaluation, running headless GPU evaluations with EGL renderi...

GrayCodeAI/starling · 40 tokens

dse-loop

Autonomous design space exploration loop for computer architecture and EDA. Runs a program, analyzes results, tunes parameters, and iterates until objective is met or timeout. Use when user says "DSE", "design space exploration", "sweep parameters", "optimize", "find best config", or wants iterative parameter tuning.

wanshuiyin/Auto-claude-code-research-in-sleep · 70 tokens

geometry

Pure-math 3D geometry toolbox — back-project masks and depth to point clouds, DBSCAN-filter noise, fit oriented bounding boxes, derive top-down/front grasp poses, and reconstruct collision worlds from RGB-D frames. Use when a workflow needs perception geometry or planner inputs computed on CPU with no model weights.

graph-robots/open-robot-skills · 64 tokens

molmoact-libero

Run the MolmoAct LIBERO checkpoint (allenai/MolmoAct-7B-D-LIBERO-0812) as a closed-loop VLA policy for the dexterous pick-and-place segment of a task. Drives a Franka Panda in the LIBERO/robosuite OSCPOSE action space from agentview + wrist cameras, served behind a vLLM-style script speaking the openpi websocket…

graph-robots/open-robot-skills · 226 tokens

ai-for-science-ai4s-main

A routing skill for AI-for-science work on Huawei Ascend NPUs. It decides whether a request belongs to performance measurement, model migration, or TensorFlow framework work, then sends it to the relevant workflow.

ascend-ai-coding/awesome-ascend-skills · 78 tokens

diansai-2026

A full workflow guide for control-focused electronics competitions, covering the path from understanding the problem to testing and writing the design report.

gugugugg/diansai-2026 · 169 tokens