mcp-chat 0.2.8

Simple CLI MCP Client to quickly test and explore MCP servers from the command line

Simple CLI MCP Client Using LangChain / Python

Quickly test and explore MCP servers from the command line!

A simple, text-based CLI client for Model Context Protocol (MCP) servers built with LangChain and Python.
Suitable for testing MCP servers, exploring their capabilities, and prototyping integrations.

Internally it uses LangChain ReAct Agent and a utility function convert_mcp_to_langchain_tools() from langchain_mcp_tools.

A TypeScript equivalent of this utility is available here

Prerequisites

• Python 3.11+
• [optional] uv (uvx) installed to run Python package-based MCP servers
• [optional] npm 7+ (npx) to run Node.js package-based MCP servers
• LLM API keys from OpenAI, Anthropic, and/or Google GenAI as needed

Quick Start

• Install mcp-chat tool. This can take up to a few minutes to complete:

  pip install mcp-chat
  

• Configure LLM and MCP Servers settings via the configuration file, llm_mcp_config.json5

  code llm_mcp_config.json5
  

  The following is a simple configuration for quick testing:

  {
    "llm": {
      "model_provider": "openai",
      "model": "gpt-4o-mini",
      // "model_provider": "anthropic",
      // "model": "claude-3-5-haiku-latest",
      // "model_provider": "google_genai",
      // "model": "gemini-2.0-flash",
    },
  
    "mcp_servers": {
      "weather": {
        "command": "npx", 
        "args": ["-y", "@h1deya/mcp-server-weather"]
      },
    },
  
    "example_queries": [
      "Tell me how LLMs work in a few sentences",
      "Are there any weather alerts in California?",
    ],
  }
  

• Set up API keys

  echo "ANTHROPIC_API_KEY=sk-ant-...                                       
  OPENAI_API_KEY=sk-proj-...
  GOOGLE_API_KEY=AI..." > .env
  
  code .env
  

• Run the tool

  mcp-chat
  

  By default, it reads the configuration file, llm_mcp_config.json5, from the current directory.
  Then, it applies the environment variables specified in the .env file, as well as the ones that are already defined.

Building from Source

See README_DEV.md for details.

Features

• Easy setup: Works out of the box with popular MCP servers
• Flexible configuration: JSON5 config with environment variable support
• Multiple LLM providers: OpenAI, Anthropic, Google (GenAI)
• Command & URL servers: Support for both local and remote MCP servers
• Local MCP Server logging: Save stdio MCP server logs with customizable log directory
• Interactive testing: Example queries for the convenience of repeated testing

Limitations

• Tool Return Types: Currently, only text results of tool calls are supported. It uses LangChain's response_format: 'content' (the default) internally, which only supports text strings. While MCP tools can return multiple content types (text, images, etc.), this library currently filters and uses only text content.
• MCP Features: Only MCP Tools are supported. Other MCP features like Resources, Prompts, and Sampling are not implemented.

Usage

Basic Usage

mcp-chat

By default, it reads the configuration file, llm_mcp_config.json5, from the current directory.
Then, it applies the environment variables specified in the .env file, as well as the ones that are already defined.
It outputs local MCP server logs to the current directory.

With Options

# Specify the config file to use
mcp-chat --config my-config.json5

# Store local (stdio) MCP server logs in specific directory
mcp-chat --log-dir ./logs

# Enable verbose logging
mcp-chat --verbose

# Show help
mcp-chat --help

Supported LLM Providers

• OpenAI: gpt-4o, gpt-4o-mini, etc.
• Anthropic: claude-sonnet-4-0, claude-3-5-haiku-latest, etc.
• Google (GenAI): gemini-2.0-flash, gemini-1.5-pro, etc.

Configuration

Create a llm_mcp_config.json5 file:

• The configuration file format for MCP servers follows the same structure as Claude for Desktop, with one difference: the key name mcpServers has been changed to mcp_servers to follow the snake_case convention commonly used in JSON configuration files.
• The file format is JSON5, where comments and trailing commas are allowed.
• The format is further extended to replace ${...} notations with the values of corresponding environment variables.
• Keep all the credentials and private info in the .env file and refer to them with ${...} notation as needed

{
  "llm": {
    "model_provider": "openai",
    "model": "gpt-4o-mini",
    // model: "o4-mini",
  },
  
  // "llm": {
  //   "model_provider": "anthropic",
  //   "model": "claude-3-5-haiku-latest",
  //   // "model": "claude-sonnet-4-0",
  // },

  // "llm": {
  //   "model_provider": "google_genai",
  //   "model": "gemini-2.0-flash",
  //   // "model": "gemini-2.5-pro-preview-06-05",
  // }

  "example_queries": [
    "Tell me how LLMs work in a few sentences",
    "Are there any weather alerts in California?",
    "Read the news headlines on bbc.com",
  ],

  "mcp_servers": {
    // Local MCP server that uses `npx`
    "weather": {
      "command": "npx", 
      "args": [ "-y", "@h1deya/mcp-server-weather" ]
    },

    // Another local server that uses `uvx`
    "fetch": {
      "command": "uvx",
      "args": [ "mcp-server-fetch" ]
    },

    "brave-search": {
      "command": "npx",
      "args": [ "-y", "@modelcontextprotocol/server-brave-search" ],
      "env": { "BRAVE_API_KEY": "${BRAVE_API_KEY}" }
    },

    // Remote MCP server via URL
    // Auto-detection: tries Streamable HTTP first, falls back to SSE
    "remote-mcp-server": {
      "url": "https://api.example.com/..."
    },

    // Server with authentication
    "github": {
      "type": "http",  // recommended to specify the protocol explicitly when authentication is used
      "url": "https://api.githubcopilot.com/mcp/",
      "headers": {
        "Authorization": "Bearer ${GITHUB_PERSONAL_ACCESS_TOKEN}"
      }
    }
  }
}

Environment Variables

Create a .env file for API keys:

OPENAI_API_KEY=sk-ant-...
ANTHROPIC_API_KEY=sk-proj-...
GOOGLE_API_KEY=AI...

# Other services as needed
GITHUB_PERSONAL_ACCESS_TOKEN=github_pat_...
BRAVE_API_KEY=BSA...

Popular MCP Servers to Try

There are quite a few useful MCP servers already available:

• MCP Server Listing on the Official Site

Troubleshooting

Common Issues

1. Missing API key: Make sure your .env file contains the required API key
2. Server not found: Ensure MCP server packages are available via npx
3. Permission errors: Check file permissions for log directory

Getting Help

• Check the logs in your specified log directory
• Use --verbose flag for detailed output
• Refer to MCP documentation

License

MIT License - see LICENSE file for details.

Contributing

Issues and pull requests welcome! This tool aims to make MCP server testing as simple as possible.