alterlab-molecular-dynamics

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external molecular structure and trajectory files, representing a potential surface for indirect prompt injection.
  • Ingestion points: The skill reads data from PDB, DCD, XTC, and other trajectory formats through functions like prepare_system_from_pdb and load_trajectory in SKILL.md and references/mdanalysis_analysis.md.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between data and control instructions within these files.
  • Capability inventory: The agent is granted Bash(python:*) and Bash(uv:*) tools, as well as Write and Edit permissions, allowing it to execute code and modify the file system based on its interpretation of the data.
  • Sanitization: The skill relies on the internal parsers of scientific libraries (OpenMM, MDAnalysis) for data handling; no additional sanitization or validation logic is specified in the prompt instructions.
  • [EXTERNAL_DOWNLOADS]: The skill documentation suggests installing the pdbfixer tool directly from a public Git repository (github.com/openmm/pdbfixer.git). While this targets an established scientific organization's repository, it involves fetching and installing code from a remote source at runtime.
  • [COMMAND_EXECUTION]: The skill makes extensive use of shell-based tools to run Python simulations and analysis scripts. The agent is instructed to perform complex operations, including system setup, energy minimization, and trajectory processing, via terminal commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 02:36 PM