返回 MCP 目录
public公开dns本地运行

Minds MCP

Minds MCP 服务器是一个基于模型上下文协议(MCP)的服务器,旨在通过标准化接口实现大型语言模型(LLM)与 Minds SDK 的交互。它提供了资源模板和工具模板,允许用户检索特定 Mind 数据并生成补全。服务器支持通过环境变量进行配置,如调试模式、日志级别、主机地址和端口等。

article

README

Minds MCP Server

An MCP (Model Context Protocol) server for Minds, allowing LLMs to interact with the Minds SDK through a standardized interface.

Installation

Install the required dependencies (virtual environment recommended)

python -m venv env
source env/bin/activate
pip install -r requirements.txt
pip install -r requirements-dev.txt

Configuration

The server can be configured using environment variables:

FastMCP Settings

  • FASTMCP_DEBUG: Enable debug mode (default: false)
  • FASTMCP_LOG_LEVEL: Set log level (default: INFO)
  • FASTMCP_HOST: Host to bind to (default: 0.0.0.0)
  • FASTMCP_PORT: Port to bind to (default: 8000)
  • FASTMCP_SSE_PATH: Path for SSE events (default: /sse)
  • FASTMCP_MESSAGE_PATH: Path for messages (default: /messages/)

Minds Settings

  • MINDS_BASE_URL: Base URL for the Minds API (default: https://mdb.ai)

Usage

Starting the Server

python -m server

Resource Templates

The server exposes the following resource templates according to the MCP specification:

  • minds://{api_key} - List all available Minds
  • minds://{mind_name}/{api_key} - Get a specific Mind by name

Tool Templates

The server provides the following tool templates:

  • completion - Generate a completion using a specified Mind
    • Parameters:
      • mind_name: The ID of the Mind to use
      • message: The message to complete
      • api_key: The Minds API key

Development

Running Tests

# Run all tests
python -m pytest

# Run only unit tests
python -m pytest tests/unit

Development Mode

For easier development, you can use the MCP development server:

mcp dev server.py

This will start the server and open the MCP Inspector in your browser for testing.

MCP Protocol

This server implements the Model Context Protocol (MCP), which provides:

  • Resource Templates: Parameterized URL patterns for retrieving data
  • Tool Templates: Function signatures for executing actions
  • Prompt Templates: Reusable interaction templates (not used in this server)

MCP clients can:

  1. Discover available resource and tool templates
  2. Instantiate templates with parameters
  3. Access resources and call tools in a standardized way
help

运行方式说明

cloud

托管运行

托管运行通常表示这个 MCP Server 由服务方环境承载,用户一般按页面提供的连接方式或授权流程接入,不需要在本地长期启动一个 MCP 进程

  1. 打开服务方连接页
  2. 完成授权或复制端点
  3. 在 MCP 客户端中连接
terminal

本地运行 / 其它方式

本地运行通常需要用户在自己的电脑或服务器上安装依赖,把 server_config 复制到 MCP 客户端,并按 env_schema 补齐环境变量、密钥或其它配置

  1. 复制 server_config
  2. 安装所需依赖
  3. 补齐环境变量后重启客户端