The Agent Skills Directory

[SAFE]: The skill implements rigorous validation logic for ML planning, requiring the agent to cite official documentation for every code block and parameter used. This architecture promotes accuracy and prevents hallucinations without introducing security risks.- [SAFE]: All external URLs referenced in the skill (such as huggingface.co, openai.com, anthropic.com, and vllm.ai) are official documentation sites for well-known and trusted technology services.- [SAFE]: No indicators of data exfiltration, credential theft, or unauthorized persistence were found. The skill instructions specifically recommend safe practices, such as using environment variables for sensitive API keys in generated code.- [SAFE]: No obfuscation techniques, hidden commands, or malicious prompt injection patterns are present. The instructional constraints ('Hard rules') are focused on ensuring the quality and grounding of the output rather than bypassing agent safety protocols.

ml-plan