The Agent Skills Directory

[SAFE]: The skill follows security best practices for handling API credentials by using environment variables and .env files, avoiding hardcoded secrets.
[SAFE]: File system operations in scripts/banner_creator.py are protected by path validation and filename sanitization (sanitize_filename) to prevent path traversal vulnerabilities.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as user-supplied strings (message, sub_copy, cta) are interpolated into prompts for the Gemini API.
Ingestion points: CLI arguments passed to build_banner_prompt and generate_copy_text in scripts/banner_creator.py.
Boundary markers: Absent; the skill uses raw string interpolation or simple double-quote wrapping.
Capability inventory: Performs network requests (Gemini API) and file system writes (saving images and markdown files).
Sanitization: No content validation or instruction-filtering is implemented for the prompt inputs.
[EXTERNAL_DOWNLOADS]: The download_reference_image function in scripts/banner_creator.py uses the requests library to fetch content from user-supplied URLs. While this is a functional requirement for reference images, it creates a potential SSRF (Server-Side Request Forgery) surface.

banner-creator