auto-mcp-client

What is auto-mcp-client?

auto-mcp-client is a Java project based on Spring AI 1.0.0-M7 that encapsulates the mcp-client service. It supports communication with MCP Servers via stdio and HTTP SSE, and provides automatic connection management including initialization, health checks, timeout closing, and connection reuse. It is designed for Java developers building multi-agent or LLM-integrated applications.

How to use auto-mcp-client?

Ensure JDK 17+ and Maven/Gradle are installed. Configure bootstrap.yml with Nacos configuration center and import the mcp-client-dev.yml config. Modify the built-in default-mcp-registry.json to set your own API keys and MCP server addresses. Also set the node service path in the configuration file. The default model is Alibaba Bailian’s qwen2.5-72b-instruct (OpenAI protocol), but you can switch to a private OpenAI-compatible model.

Key features of auto-mcp-client

• Multi-protocol support: stdio and HTTP SSE
• Automatic connection initialization on startup
• Automatic health checks for connection availability
• Automatic timeout closing to release idle connections
• Connection reuse within timeout windows
• Environment isolation between multiple agents using the same MCP server

Use cases of auto-mcp-client

• Compare weather between cities and generate bar charts
• Analyze database data (e.g., Fujian GDP top 5) and create pie charts
• Extract content from web pages and summarize into Markdown

FAQ from auto-mcp-client

What models does auto-mcp-client support?

It supports Alibaba Bailian’s qwen2.5-72b-instruct by default (OpenAI protocol) and can be configured to use any private OpenAI-compatible model.

How does it connect to MCP servers?

It uses stdio (local subprocess) or HTTP SSE (Server-Sent Events) to communicate with MCP servers.

Is there automatic reconnection or health monitoring?

Yes, the client periodically checks connection health and automatically reinitializes or closes connections as needed.

Is pricing mentioned?

No, the README does not discuss pricing or licensing.

Are there any known limitations?

The README does not list any specific limitations.