dm_control
DeepMind's MuJoCo-based physics simulation library and continuous-control RL benchmark suite.
dm_control
dm_control is Google DeepMind's software stack for physics-based simulation and reinforcement learning research. It wraps the MuJoCo physics engine with a clean Python API and provides a comprehensive suite of continuous-control tasks (the DeepMind Control Suite) that have become a standard benchmark in RL research.
Key Features
- Python bindings for the full MuJoCo C API via
dm_control.mujoco - DeepMind Control Suite: 28 physics tasks across locomotion, manipulation, and more
- Composer framework for procedurally assembling complex tasks from reusable components
- Locomotion arena library with varied terrains and procedural mazes
- Pixel and state observations, off-screen rendering, and deterministic resets
Quick Start
pip install dm-control
from dm_control import suite
import numpy as np
env = suite.load("cheetah", "run")
action_spec = env.action_spec()
timestep = env.reset()
while not timestep.last():
action = np.random.uniform(action_spec.minimum, action_spec.maximum)
timestep = env.step(action)
print(timestep.reward)
npx ai-supply add dm-control-physics-rl
Curated mirror of the open-source dm_control (Apache-2.0). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/dm-control-physics-rl/check). Click a policy:
Consume dm_control programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/dm-control-physics-rl
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/dm-control-physics-rl/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add dm-control-physics-rl
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/dm-control-physics-rl/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "dm-control-physics-rl" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.