github-mcp-server

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to read and act on user-generated content from public GitHub (e.g., "Show me the contents of main.go", "Show open issues", "Review PR #234"), so the skill fetches and interprets untrusted third‑party repository/issue/PR content from GitHub.

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

• Potentially malicious external URL detected (high risk: 0.90). I flagged https://api.githubcopilot.com/mcp/ (used at runtime as the MCP endpoint that serves repository files which are injected into the model context and can directly influence prompts) and ghcr.io/github/github-mcp-server (pulled/executed via Docker in the local-server path, which fetches and runs remote code required for the skill).