arthas_mcp_server

Arthas MCP Server provides a Python-based diagnostic toolkit that integrates with Alibaba Arthas to help AI assistants analyze Java applications. The server exposes a set of tools for connecting to Arthas, fetching JVM and thread information, inspecting memory usage, executing Arthas commands, and analyzing performance. It is designed to be invoked via the MCP workflow and is suitable for LLM-assisted diagnosis and runtime operations on Java processes. You can connect to Arthas WebConsole, retrieve JVM state, and perform dynamic analysis tasks while the host application runs. The server aims to provide structured, diagnosable outputs that simplify AI-assisted debugging and performance tuning.

Prerequisites:

• Python 3.8+ (or newer)
• Java Runtime Environment (for Arthas compatibility)
• MCP tooling installed (uv/uvx) as described by the MCP provider

1. Install Python dependencies (if any are listed in requirements):

   • python -m pip install --upgrade pip
   • python -m pip install -r requirements.txt (if a requirements file exists)

2. Install the MCP client tooling (uv/uvx) according to your environment documentation. Example (illustrative):

   • Install uv/uvx following the project’s installation guide. This often involves installing via pipx or a package manager as directed by MCP docs.

3. Prepare the Arthas MCP server files:

   • Ensure you have the Arthas MCP server directory structure as shown in the repository (src/, main.py, etc.).
   • Place this Arthas MCP server in a local directory and note the path for the MCP config.

4. Configure the MCP server in your environment:

   • Use the provided mcp_config example (adjust paths and environment variables to your setup).

5. Run the server:

   • Depending on your setup, start via the MCP CLI with the corresponding command (for uv-based execution use the configuration shown in mcp_config).

6. Validate:

   • Check that Arthas WebConsole is reachable at ARTHAS_URL (default http://localhost:8563).
   • Use MCP tooling to connect and verify that the server reports JVM, thread, and memory info correctly.

Tips and common issues:

• Ensure Arthas is accessible from the host running the MCP server (network/port exposure).
• If ARTHAS_URL changes, update the environment variable accordingly in the mcp_config.
• The provided example uses a Windows path syntax; adapt to your OS as needed.
• If you modify main.py, restart the MCP server to apply changes.
• For development workflows, you can use uv sync and uv sync --extra dev to fetch and test updates.