The Agent Skills Directory

[SAFE]: The skill serves as a reference for prompt engineering patterns. All code samples are educational or utility-focused, demonstrating standard integration with libraries like LangChain and Pydantic.
[INDIRECT_PROMPT_INJECTION]: Several templates and the prompt optimization script interpolate variables into LLM prompts. While this introduces a theoretical surface for indirect prompt injection, it is the fundamental purpose of the skill and is implemented using standard practices.
Ingestion points: kwargs in references/prompt-templates.md, test_case.input in scripts/optimize-prompt.py.
Boundary markers: Generally absent in the provided templates, though the documentation encourages specific instructions.
Capability inventory: LLM invocations using ainvoke and complete methods.
Sanitization: No explicit sanitization of interpolated content is performed in the sample code.
[DYNAMIC_EXECUTION]: The ConditionalTemplate class in references/prompt-templates.md implements a custom logic layer using regular expressions to process conditional blocks and loops within prompts. This is a common and safe pattern for template management.

prompt-engineering-patterns