The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/imiva_mcp.py script utilizes subprocess.Popen to call the npx command. This is used to run the platform's content generation tool. The implementation uses argument lists to avoid shell injection vulnerabilities.\n- [EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and execute the @infimind/ecom-content-cli package from the NPM registry. This follows standard practices for accessing remote MCP capabilities from well-known package registries.\n- [REMOTE_CODE_EXECUTION]: The skill invokes the downloaded NPM package to perform its tasks. As this package is the official tool for the service described and is retrieved from a well-known registry, this execution is considered part of the skill's intended functionality.\n- [DATA_EXFILTRATION]: The skill sends product material and configuration to the IMIVA service at imiva.ecpro.com. This is the core functional behavior of the skill and is secured via a user-managed API token.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external product information which could potentially contain malicious instructions.\n
Ingestion points: Input parameters for the create_product_main_image_task tool defined in SKILL.md and handled by scripts/imiva_mcp.py.\n
Boundary markers: The instructions include explicit steps for the agent to verify product facts and selling points before generation.\n
Capability inventory: The skill can read local file paths for materials and perform network requests to the IMIVA API.\n
Sanitization: External inputs are encapsulated in a structured JSON-RPC format before being passed to the execution engine.

ai-ecommerce-expert-lazada-ecommerce-content