The Agent Skills Directory

[SAFE]: The skill is primarily informational and provides high-quality documentation for prompt engineering best practices. It includes reference code for common architectural patterns such as Tree-of-Thoughts and ReAct, using established development practices.
[INDIRECT_PROMPT_INJECTION]: The skill provides templates and methods for processing untrusted user input within LLM prompts. However, it explicitly advocates for security best practices to mitigate these risks, including the use of XML tags as boundary markers and the implementation of validation schemas to verify input before it is interpolated into prompts.
[DYNAMIC_EXECUTION]: The skill includes Python reference implementations for a prompt templating engine that uses standard string formatting and regular expressions to render content at runtime. This is a common and legitimate pattern for dynamic prompt generation.
[EXTERNAL_DOWNLOADS]: The documentation provides code examples that demonstrate integration with official services from established providers such as Anthropic and OpenAI, as well as well-known data science packages.

prompt-engineering