The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Python scripts (ruiguan_image_compliance_search.py and upload_image.py) designed to be executed by the agent to perform similarity searches and handle image uploads via the vendor's API.
[EXTERNAL_DOWNLOADS]: The skill interacts with external vendor endpoints for its core functionality, including the policy detection API at NEXSCOPE_PROXY_BASE and a feedback endpoint at skill-api.nexscope.com.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied image URLs and retrieves data (product titles and images) from a remote database.
Ingestion points: The imageUrl parameter in SKILL.md and processed via scripts/ruiguan_image_compliance_search.py.
Boundary markers: None explicitly defined in instructions or scripts.
Capability inventory: Python scripts perform network requests and local file writes to the user's filesystem.
Sanitization: Basic JSON structure validation is performed in the scripts.
[SAFE]: The Python scripts access the NEXSCOPE_API_KEY and NEXSCOPE_PROXY_BASE environment variables to authenticate legitimate API requests.
[SAFE]: The skill writes session logs and cached results to a nexscope directory within the user's home or current working directory to improve performance and provide persistence for detection results.

ecommerce-ruiguan-gun-parts-search