Skills
skills/swn94/claude-scientific-skills/stable-baselines3/Gen Agent Trust Hub

stable-baselines3

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [Remote Code Execution] (MEDIUM): The templates in scripts/evaluate_agent.py and scripts/train_rl_agent.py use loading functions from the stable_baselines3 library. These functions use the Python pickle module to load saved model states. Pickle is inherently insecure and can be used to execute arbitrary code if a model from an untrusted source is loaded.
  • [Command Execution] (LOW): The scripts/train_rl_agent.py script utilizes SubprocVecEnv to parallelize environment instances. This involves spawning and managing multiple Python subprocesses, which is a standard but potentially powerful capability for process management.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:48 PM