The Agent Skills Directory

[SAFE]: The skill contains purely technical documentation, architectural patterns, and validation rules for Reinforcement Learning algorithms and alignment processes.
[SAFE]: Code snippets provided in the reference files are educational examples illustrating common algorithmic pitfalls (like reward hacking or catastrophic collapse) and their legitimate solutions.
[SAFE]: No malicious patterns such as prompt injection, data exfiltration, obfuscation, or unauthorized command execution were detected across any of the files.

reinforcement-learning