The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx -y to download and execute the @infimind/ecom-content-cli package from the npm registry. This occurs at runtime to ensure the latest version of the vendor's official tool is used.
[COMMAND_EXECUTION]: The Python helper script scripts/imiva_mcp.py uses subprocess.Popen to execute the npx command. The implementation is safe as it passes command arguments as a list rather than a shell string, and interacts with the tool via stdin to prevent shell injection vulnerabilities.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection.
Ingestion points: The skill ingests untrusted product data, including "commodity material" and "real selling points," as described in SKILL.md.
Boundary markers: The instructions do not define clear delimiters or "ignore instructions" warnings for the agent when processing these external inputs.
Capability inventory: The skill has the capability to execute shell commands and interact with external APIs via scripts/imiva_mcp.py and the create_video_generation_task tool.
Sanitization: No explicit sanitization or filtering of the input data is performed in the wrapper script before processing.

ai-ecommerce-expert-kuaishou-ecommerce-content