The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data which could potentially contain malicious instructions aimed at the agent's logic.
Ingestion points: User-provided text prompts and reference images (via URL or Base64 data) are ingested as inputs for the API requests described in SKILL.md.
Boundary markers: No delimiters or specific 'ignore' instructions are used to separate user-provided content from the agent's execution script.
Capability inventory: The skill utilizes network operations (POST requests) to gateway.pixazo.ai and performs local file system operations (writing to ~/.pixazo/api-key).
Sanitization: There is no evidence of sanitization or validation of the user's prompt or image data before it is interpolated into the API request body.
[COMMAND_EXECUTION]: The skill instructs the agent to use chmod 600 when saving the API key to ~/.pixazo/api-key. This is a recommended security measure to ensure the credential file is only readable by the owner.
[EXTERNAL_DOWNLOADS]: The skill fetches model documentation from https://www.pixazo.ai/models/gfpgan.md, which is an official resource provided by the vendor.

gfpgan