The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The script scripts/imiva_mcp.py uses npx -y to download the @infimind/ecom-content-cli@latest package from the npm registry. This occurs at runtime without version pinning or integrity checks.
[REMOTE_CODE_EXECUTION]: The skill executes code downloaded dynamically from the npm registry. Using the @latest tag means the code being executed can change without user intervention, presenting a risk if the remote package is compromised.
[COMMAND_EXECUTION]: The scripts/imiva_mcp.py file utilizes subprocess.Popen to launch the npx command and the external Node.js package. This grants the remote code execution capabilities within the local environment.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from product images and descriptions which could be manipulated to influence agent behavior.
- Ingestion points: referenceImage, productImages, and subjectDescriptions parameters defined in SKILL.md.
- Boundary markers: None identified; the instructions do not implement delimiters to isolate user-provided content from instructions.
- Capability inventory: The skill can execute shell commands via the subprocess module in scripts/imiva_mcp.py.
- Sanitization: The script does not appear to perform validation or escaping of the input strings before passing them to the external CLI tool.

ai-ecommerce-expert-cross-category-viral-image