The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/imiva_mcp.py executes an external, unpinned NPM package at runtime. It uses subprocess.Popen to run npx -y @infimind/ecom-content-cli@latest, which downloads and executes the latest version of the code from the NPM registry without version pinning or integrity verification.
[COMMAND_EXECUTION]: The skill uses the Python subprocess module in scripts/imiva_mcp.py to launch shell commands. Specifically, it launches npx to initialize and communicate with a Model Context Protocol (MCP) server.
[EXTERNAL_DOWNLOADS]: The skill is designed to download and run the @infimind/ecom-content-cli package from the public NPM registry during execution, as specified in both the documentation and the helper script.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions.
Ingestion points: The skill ingests user-provided product materials including productImages, productCategory, coreSellingPoints, and usageScenario as documented in SKILL.md and handled by scripts/imiva_mcp.py.
Boundary markers: There are no explicit delimiters or instructions to the agent to treat this specific data as untrusted or to ignore embedded instructions within the materials.
Capability inventory: The skill possesses the capability to execute commands (subprocess.Popen) and perform network requests through the bridged MCP server.
Sanitization: The implementation lacks logic to sanitize, validate, or filter user-provided text or image metadata before it is processed by the agent.

ai-ecommerce-expert-pinduoduo-ecommerce-content