The Agent Skills Directory

[SAFE]: The skill uses localized Python scripts for prompt rendering and verification. Network activity is limited to downloading assets from Netlify (a well-known service) with mandatory SHA-256 integrity checks. No sensitive file access or hardcoded credentials were found.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied subjects and variables through scripts/render_prompt.py. Ingestion points include CLI arguments (--subject, --text, --var) interpolated into templates in STYLES.md. While this creates an injection surface, the skill implements a protective layer using regex patterns (STYLE_INJECTION_PATTERNS) and keyword filtering to prevent users from overriding internal style constraints. Capability is limited to text prompt generation. Boundary markers are not explicitly used in the output string, but the regex sanitization serves as an intentional defensive measure.

hand-drawn