Overview
cockroachdb-mcp-server is a production-grade, spec-aligned MCP server implemented in Python with FastAPI and CockroachDB. It provides full MCP functionality with CRUD APIs to manage model contexts, and stores context definitions as JSONB to support arbitrary input/output schemas. The server is designed to work seamlessly with the cockroachdb-mcp-client CLI for managing contexts, simulating LLM runs, exporting data, and batch testing across providers. It exposes a REST API at /contexts with endpoints to create, list, retrieve, update, and delete contexts, and includes schema bootstrapping capabilities that can be triggered via a CLI flag (--init-schema) or an environment variable (MCP_AUTO_INIT_SCHEMA). The bootstrap creates a mcp_contexts table with id UUID, context_name, context_version, body JSONB, and created_at fields. The project emphasizes interoperability with CockroachDB, URL auto-detection and dialect fixes, structured logging, and readiness for extensions such as /run, /deploy, and /evaluate.
Features
REST API for MCP context management
Exposes full CRUD APIs for managing MCP contexts via the /contexts endpoints (create, list, get, update, delete).
Schema bootstrapping
Supports automatic and CLI/ENV-driven schema initialization to create the mcp_contexts table.
CRDB URL auto-detection and dialect fix
Automatically rewrites CRDB URLs for compatibility and fixes dialect issues.
JSONB-based context storage
Stores context definitions as JSONB to support arbitrary input/output schemas.
Structured logging
Provides structured logging with configurable log levels for observability.
LLM-ready CLI integration
Works with the cockroachdb-mcp-client CLI for context management, simulations, exports, and batch tests.
Extensible MCP workflow endpoints
Ready for future /run, /deploy, /evaluate extensions to support end-to-end MCP workflows.
Who Is This For?
- ML Engineers:Manage MCP contexts via CRUD REST APIs and run LLM simulations using the CLI.
- Backend / Full-stack Developers:Build integrations with FastAPI services and store MCP contexts in JSONB within CockroachDB.
- DevOps / Platform Engineers:Configure schema bootstrapping, manage DB connections via CRDB URLs, and enable observability with structured logging.
