The Agent Skills Directory

[SAFE]: The skill provides documentation and code templates for Stable Baselines3, a well-known library for reinforcement learning. No malicious patterns such as prompt injection, data exfiltration, or obfuscation were detected.
[EXTERNAL_DOWNLOADS]: The skill recommends installing standard Python packages (stable-baselines3, gymnasium) using standard package managers. These are well-known libraries in the AI/ML community.
[COMMAND_EXECUTION]: The provided scripts (train_rl_agent.py, evaluate_agent.py) perform local file system operations like os.makedirs to manage logs and model checkpoints, which is standard behavior for machine learning workflows and presents no security risk.
[DATA_EXFILTRATION]: No evidence of unauthorized network activity or exfiltration of sensitive files was found. Network activity is limited to optional TensorBoard logging to a local directory.

stable-baselines3