MCP Tools & Marketplace

LingXizhi implements the Model Context Protocol (protocol version 2025-06-18, compatible with 2024-11-05), supports all three transports — stdio / SSE / Streamable HTTP — and supports OAuth authorization flows. Any MCP Server that conforms to the protocol can be connected to give the AI additional tools.

Installation & Configuration

Declare MCP servers in .lxz/mcp.json at the project root:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/tmp"]
    }
  }
}

Once configured, the tools provided by the server are automatically registered in the AI's tool list, and call records can be viewed in the Tools tab of the AI panel.

Tool Marketplace

The tool marketplace lets you browse officially curated MCP servers, with entries showing category, download count, and rating — one-click install writes the configuration automatically. Official presets include:

• filesystem: file system operations — read/write files, list directories, search files.
• github: GitHub integration — create and manage issues and PRs, search code.
• gitlab: GitLab integration — project management, merge requests, and issues.

Build Your Own MCP Server

Implement the standard MCP methods (initialize, tools/list, tools/call) and you can connect:

1. Choose a transport: use stdio for local processes (subprocess standard I/O, communicating in Content-Length frames); use SSE or Streamable HTTP for remote services.
2. Write your service's start command into mcpServers in .lxz/mcp.json.
3. Your tools can then be called directly in conversation, with call records shown in the Tools tab.