claude-code-mcp

MCP servers enable Claude Code to access PostgreSQL databases through configured server templates. Set up a PostgreSQL MCP server in your MCP.json configuration, bind connection credentials via environment variables, and validate connectivity through the CLI. This establishes secure, allowlisted database queries from Claude Code.

Available MCP server templates include PostgreSQL for database access, filesystem for file operations, Git for version control integration, Brave Search for web queries, and Slack for messaging. Each template provides preconfigured server definitions you can customize and deploy via CLI or JSON configuration.

Yes, MCP supports multi-server orchestration. Configure multiple MCP servers in a single JSON file, each with distinct transports—HTTP, stdio, or SSE—and environment-based access controls. Claude Code loads and manages all configured servers simultaneously, allowing unified access to databases, filesystems, APIs, and external services.

MCP servers enforce security through environment-based access controls, allowlisted filesystem paths, and server-specific capability restrictions. Bind sensitive credentials—database passwords, API keys—via environment variables rather than hardcoding them in configuration, and define granular permissions for each server's operations.

Yes, the extensible driver schema supports custom MCP servers for proprietary data sources and APIs. Build and deploy custom servers using the provided framework, configure them in MCP.json with appropriate transports and environment variables, and integrate them alongside built-in templates for complete data source coverage.

MCP servers support three transport protocols: HTTP for remote connections, stdio for local process communication, and SSE (Server-Sent Events) for streaming updates. Select the transport that matches your server's architecture and deployment environment in the MCP.json configuration.