The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx to download and execute the @infimind/ecom-content-cli@latest package from the npm registry during execution. This represents a dependency on a remote, unpinned package from a third-party source.
[COMMAND_EXECUTION]: The helper script scripts/imiva_mcp.py uses subprocess.Popen to launch the downloaded Node.js CLI tool, facilitating communication between the local environment and the IMIVA MCP server.
[PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by ingesting and processing untrusted user data which is subsequently passed to a remote AI service.
Ingestion points: Data enters the skill via the --args and --args-file parameters in scripts/imiva_mcp.py. The script also has the capability to read arbitrary local JSON files provided via the file path argument.
Boundary markers: There are no explicit delimiters or protective instructions in the skill's scripts to differentiate user-supplied content from system instructions.
Capability inventory: The skill can execute various ecommerce tasks including create_smart_refine_task, which processes images and prompts, and get_user_tasks to query task history.
Sanitization: While the script validates that inputs are valid JSON objects, it does not perform any sanitization or filtering of the textual content within those objects before transmission to the remote API.

ai-ecommerce-expert-product-material-detail-image