build-mcp

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to use WebFetch and web search to load public URLs (e.g., https://modelcontextprotocol.io/llms-full.txt and raw.githubusercontent.com links) as required steps in Phase 1, meaning the agent will ingest untrusted, user-accessible third-party content that can materially influence tool design and subsequent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs using WebFetch at runtime to load https://modelcontextprotocol.io/llms-full.txt (and the SDK READMEs at https://raw.githubusercontent.com/modelcontextprotocol/python-sdk/main/README.md and https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/main/README.md), which the agent must fetch and then use as authoritative protocol/specification guidance that directly controls its prompts/behavior and is treated as a required dependency.