claude-code-mcp-manual-install
Warn
Audited by Snyk on Feb 16, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.80). The skill includes instructions to add HTTP MCP servers by setting an arbitrary "url" in ~/.claude.json (see the "HTTP タイプ" example with "https://example.com/mcp") and notes that MCP servers are loaded at session start, so the agent could fetch and interpret content from arbitrary third‑party web endpoints.
Audit Metadata