The Agent Skills Directory

[SAFE]: The skill uses standard Python libraries like Pillow, requests, and numpy to handle image processing and network communication. The scripts provide clear input validation and use well-defined API endpoints (using example.com as placeholders).
[COMMAND_EXECUTION]: The skill defines a system dependency in its metadata that executes a benign echo command during the loading process. This is used for logging purposes and poses no security risk.
[EXTERNAL_DOWNLOADS]: The multimedia generation scripts (generate_image.py, generate_audio.py) are designed to download generated assets from remote API endpoints. This is a functional requirement of the skill and does not involve the execution of untrusted code.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it interpolates user-supplied text (such as product names and descriptions) into prompts sent to generative AI models.
Ingestion points: User-provided strings for product scripts in assets/templates/product-recommendation-script.md.
Boundary markers: None identified.
Capability inventory: The skill can perform network requests to external APIs via requests.post.
Sanitization: No specific sanitization or filtering of user input is performed before interpolation. However, this behavior is essential to the skill's primary purpose and is considered a low-risk design pattern.

agentkit-multimedia-shopping