Scout Monitoring's MCP gives your AI assistant direct access to production performance data through Scout's API. You get tools to pull traces with line-level profiling, query error groups with full backtraces, and fetch insights on N+1 queries, memory bloat, and slow endpoints across Rails, Django, FastAPI, Laravel, and other supported frameworks. The server exposes operations like get_app_trace for detailed execution spans, get_app_insights for performance problems, and get_app_error_groups for recent failures. It's designed for having your AI open issues, generate fix PRs, or debug performance problems without leaving your editor. Runs via Docker with your Scout API key, and includes a setup wizard for Cursor, Claude Desktop, and VS Code Copilot.
claude mcp add --transport stdio scoutapp-scout-mcp-local uvx scout-mcp-local