mcp-builder

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly instructs the agent to WebFetch external public URLs (e.g., https://modelcontextprotocol.io/llms-full.txt and https://raw.githubusercontent.com/modelcontextprotocol/.../README.md), causing the agent to load and interpret third-party web content (including user-maintained GitHub READMEs) which could contain untrusted or injected instructions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent at runtime to WebFetch and load external plain-text/spec files (https://modelcontextprotocol.io/llms-full.txt, https://raw.githubusercontent.com/modelcontextprotocol/python-sdk/main/README.md, https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/main/README.md), which would be fetched and injected into the agent's context to directly control prompts/instructions.