read-github

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests documentation and external URLs from public GitHub repositories via the gitmcp.io MCP service (see SKILL.md workflow steps "fetch documentation" and "fetch-url" and scripts/gitmcp.py which calls tools like fetch_{repo}_documentation and fetch_generic_url_content), so untrusted third‑party content can be read and materially influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The script runs npx to execute the remote "mcp-remote" client against converted gitmcp.io URLs (e.g. https://gitmcp.io/owner/repo), which at runtime fetches/executes external npm code and exchanges JSON‑RPC tool responses that can directly supply instructions/content to the agent.