The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill accepts and processes untrusted data from users which is then interpolated into API requests for image processing.
Ingestion points: User-provided image descriptions and external image URLs (image_url) specified in SKILL.md.
Boundary markers: None identified in the skill instructions to delimit untrusted content.
Capability inventory: The skill uses the requests library in Python and fetch in Node.js to perform network operations to the Pixazo API.
Sanitization: There is no evidence of input validation or sanitization before the data is sent to the API gateway.
[EXTERNAL_DOWNLOADS]: The skill is configured to fetch additional documentation and parameter specifications from the vendor's official website.
Evidence: Fetches configuration and full reference data from https://www.pixazo.ai/models/smart-resize.md.
[COMMAND_EXECUTION]: The documentation mentions a command for managing multiple skills via a package runner.
Evidence: npx skills add Pixazo-AI/skills --skill '*' is referenced in the related skills section.

smart-resize