project-setup-info-local

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

• Third-party content exposure detected (high risk: 1.00). The mcp-server rules explicitly instruct the agent to "visit https://github.com/modelcontextprotocol" and to use the fetch_webpage tool to retrieve and act on content from the public URL https://modelcontextprotocol.io/llms-full.txt, which is untrusted third-party content that the agent must read and that can change its 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 calls the fetch_webpage tool at runtime to load https://modelcontextprotocol.io/llms-full.txt, and that fetched content is required and would directly supply implementation instructions that control the agent's behavior.