content-safety-api
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of user-provided image URLs and base64-encoded image data, creating a potential surface for indirect injection if the processed content contains malicious instructions.
- Ingestion points:
image_url, base64 data URLs, and user prompts. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined for the input data.
- Capability inventory: File writing (
~/.pixazo/api-key) and network operations viacurl,requests, andfetchtogateway.pixazo.ai. - Sanitization: The instructions do not specify validation or sanitization of the input URLs before they are passed to the API.
- [EXTERNAL_DOWNLOADS]: The skill is designed to fetch additional model reference information from the vendor's official website.
- Evidence: Instructions direct the agent to fetch documentation from
https://www.pixazo.ai/models/content-safety-api.mdto resolve parameter requirements at runtime. - [COMMAND_EXECUTION]: The documentation includes a command for installing related skills using
npxfrom the vendor's repository. - Evidence: The skill references
npx skills add Pixazo-AI/skills --skill '*'as a method for expanding agent capabilities within the vendor ecosystem.
Audit Metadata