The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The helper script scripts/imiva_mcp.py invokes npx -y @infimind/ecom-content-cli@latest. This pattern downloads and executes the package binary directly from the npm registry every time the skill is used, which constitutes a form of remote code execution.
[COMMAND_EXECUTION]: The skill executes shell commands using subprocess.Popen to manage the Node.js process and the MCP (Model Context Protocol) tool. While the implementation uses argument lists to mitigate shell injection risks, it enables the execution of processes defined by an external package.
[EXTERNAL_DOWNLOADS]: The skill depends on the @infimind/ecom-content-cli package, which is fetched from an external source at runtime during initialization.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection through user-provided inputs.
Ingestion points: Data enters via parameters like referenceImage, productImages, and subjectDescriptions in the create_visual_migration_task tool.
Boundary markers: The skill lacks explicit boundary markers or instructions for the agent to ignore instructions embedded in these fields.
Capability inventory: The skill can execute shell commands and perform network operations through the external service.
Sanitization: There is no evidence of input validation or escaping performed on user-supplied strings before processing.

ai-ecommerce-expert-bestseller-main-image