The Agent Skills Directory

[SAFE]: The skill is primarily educational and provides utility scripts for managing and optimizing Large Language Model (LLM) prompts. No malicious code or harmful instructions were found during the analysis.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides templates that interpolate external data (such as user queries or RAG context) into prompts. While this is an inherent risk in prompt engineering, the skill encourages best practices such as specific role definitions and structural delimiters, which help mitigate these risks.\n- [DYNAMIC_EXECUTION]: The template systems described in the references use standard string manipulation and regex-based logic. There is no usage of unsafe execution functions like eval() or exec() for processing external content.\n- [EXTERNAL_DOWNLOADS]: The code snippets and scripts reference well-known, legitimate Python libraries including openai, sentence-transformers, scikit-learn, and numpy. No suspicious or unverified dependencies are requested.

prompt-engineering-patterns