mujoco-audit

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses uv run to execute its audit suite, which involves downloading standard scientific Python dependencies (such as mujoco, trimesh, and numpy) from the official Python Package Index (PyPI).
  • [COMMAND_EXECUTION]: The instructions guide the agent to perform asset diagnostics and remediation by executing a local Python script and managing files within the artifact directory.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external MuJoCo MJCF assets and geometry files provided by the user. 1. Ingestion points: The agent loads and parses MJCF XML files, .mjz archives, and .obj mesh files. 2. Boundary markers: There are no explicit prompt-level delimiters defined for the data content; parsing is handled by standard libraries. 3. Capability inventory: Capabilities include reading and writing files within the workspace and executing a subprocess for the audit suite. 4. Sanitization: The audit script includes built-in hygiene checks to validate MJCF standards and mesh topology.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 02:30 PM