The Agent Skills Directory

[SAFE]: The skill consists of 15 markdown files serving as a knowledge base for reinforcement learning algorithms. A thorough scan across all 10 threat categories confirms the absence of malicious patterns. The Python code provided is instructional pseudocode or standard implementation of RL agents (e.g., DQN, PPO, SAC) and diagnostic tools. There are no executable scripts that perform network operations, access sensitive file paths, or attempt privilege escalation. The skill does not define any tools, minimizing the attack surface for indirect prompt injection. All references to external packages and environment APIs (Gymnasium) align with legitimate, well-known software in the machine learning ecosystem.

using-deep-rl