The Agent Skills Directory

[Indirect Prompt Injection] (SAFE): The skill provides a library of templates and an optimization script (scripts/optimize-prompt.py) that use variable interpolation (e.g., {text}, {user_query}). This represents the standard functional surface for prompt engineering. While no explicit sanitization or boundary markers are defined in the templates, this is expected for a developer-focused toolkit.
[External Downloads] (SAFE): The included Python script imports numpy for metric calculations and the documentation references the openai library. These are standard, trusted dependencies for the skill's stated purpose of prompt optimization and LLM orchestration.
[Dynamic Execution] (SAFE): While the documentation includes Python snippets for educational purposes, the skill itself does not dynamically execute untrusted code. The optimize-prompt.py script uses a controlled execution flow with a mock client for demonstration.

prompt-pro