This MCP server dashboard provides a unified management surface for multiple MCP servers. It exposes a single streaming HTTP interface (and optional SSE endpoints) that routes requests to any configured MCP server, enabling real-time interactions with tools or assistants across different domains. You can monitor status and performance from the built-in management console, enable or disable individual servers, and organize servers into groups for easier access control. Tools like the amap MCP, Playwright-based MCPs, fetch-based MCPs, and Slack- integration MCPs can all be reached through the same endpoints, simplifying integration with AI clients and automation pipelines. Advanced features include smart routing, which leverages vector embeddings and semantic search to surface the most relevant tools for a given task, and group-based endpoints that allow targeted access to subsets of servers.

Prerequisites:

• Docker and Docker Compose (for containerized deployment) or Node.js environment for local dev/testing
• Git
• Optional PostgreSQL with pgvector if you enable Smart Routing
• Basic network access on port 3000 (or as configured)

Installation steps:

1. Clone the repository: git clone https://github.com/zxerai/MCP-Server.git cd MCP-Server

2. Install dependencies (local development): pnpm install

3. Prepare your mcp_settings.json configuration (example included in the Quick Start): Create a file named mcp_settings.json with the following structure and your own keys (see README for details).

4. Run in development mode (local): pnpm dev

   or start backend/frontend separately if required

5. Run with Docker (quick start): docker-compose up -d

   or start with a specific profile (e.g., nginx) if you have reverse proxy needs

6. Run a custom Docker command with your configuration file mounted: docker run -p 3000:3000 -v ./mcp_settings.json:/app/mcp_settings.json -v ./data:/app/data mcpserver:1.0.0

7. Access the management console: Open http://localhost:3000 and log in with the default credentials (admin/admin123 by default).

Note: If you plan to enable Smart Routing, ensure PostgreSQL with pgvector is available and you have an OpenAI (or compatible) embedding service configured.

Tips and considerations:

• Default credentials should be changed before production deployment (admin/admin123).
• If you use Nginx as a reverse proxy, consider disabling buffering (proxy_buffering off) for smoother streaming responses.
• When using Docker, if port 3000 is in use, set MCPSERVER_PORT via environment variables or adjust your docker-compose definitions.
• For Smart Routing, ensure your PostgreSQL database is accessible and that pgvector is installed and configured in your environment.
• Group-based endpoints help isolate server access; remember that group name takes precedence if both group and server names collide in routing.
• Always secure tokens and secrets (JWT, API keys) and rotate credentials periodically.