The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions require the user to clone a repository from an unverified GitHub account and execute its code.
Evidence: The setup process instructs running git clone https://github.com/dhapat3927/mcp-edd-analytics-vantage.git followed by npm install and npm start.
Concern: This involves downloading and executing Node.js code from a source that is not recognized as a trusted organization, which could lead to the execution of unvetted logic.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the ingestion of untrusted data from an external e-commerce platform.
Ingestion points: Data enters the context via the get_products, get_customers, and get_orders tool calls to the WordPress REST API.
Boundary markers: No boundary markers (such as XML tags or delimiters) or instructions to ignore embedded commands are present in the documentation.
Capability inventory: The skill can access sensitive sales reports, customer lists, and license keys.
Sanitization: There is no evidence of sanitization or filtering for the data returned by the API before it is processed by the AI agent.

commercevault-edd-commerce-orchestrator