Scorable MCP Server is a bridge between the Scorable API and MCP client applications, exposing Scorable evaluators as MCP tools that allow AI assistants and agents to evaluate responses against defined quality criteria. The server fetches available evaluators from the Scorable API on startup and makes them accessible through MCP endpoints. It implements Server-Sent Events (SSE) for network deployment and is compatible with MCP clients such as Cursor. The server provides tools including list_evaluators, run_evaluation, run_evaluation_by_name, run_coding_policy_adherence, list_judges, and run_judge, enabling both single evaluations and judge-based (LLM-as-a-judge) workflows. A typical setup involves obtaining a Scorable API key and running the server via Docker, with the README showing a docker command and logs that illustrate the server initializing, fetching evaluators, and indicating that the SSE server is listening on an endpoint. The README also demonstrates usage patterns in Python for listing evaluators, executing evaluations by ID or name, and integrating with Cursor and other MCP clients. Endpoints mentioned include /mcp (preferred) and /sse (backward compatible).