MCP REST API and CLI Client

MCP REST API and CLI Client

A simple REST API and CLI client to interact with Model Context Protocol (MCP) servers.

Key Features

1. MCP-Compatible Servers

• Supports any MCP-compatible servers servers.
• Pre-configured default servers:
  • SQLite (test.db has been provided with sample products data)
  • Brave Search
• Additional MCP servers can be added in the mcp-server-config.json file

2. Integrated with LangChain

• Leverages LangChain to execute LLM prompts.
• Enables multiple MCP servers to collaborate and respond to a specific query simultaneously.

3. LLM Provider Support

• Compatible with any LLM provider that supports APIs with function capabilities.
• Examples:
  • OpenAI
  • Claude
  • Gemini
  • AWS Nova
  • Groq
  • Ollama
  • Essentially all LLM providers are supported as long as they provide a function-based API. Please refer langchain documentation for more details.

Setup

1. Clone the repository:

   git clone https://github.com/rakesh-eltropy/mcp-client.git
   

2. Navigate to the Project Directory After cloning the repository, move to the project directory:

   cd mcp-client
   

3. Set the OPENAI_API_KEY environment variable:

   export OPENAI_API_KEY=your-openai-api-key
   

   You can also set the OPENAI_API_KEY in the mcp-server-config.json file.

   You can also set the provider and model in the mcp-server-config.json file. e.g. provider can be ollama and model can be llama3.2:3b.

4.Set the BRAVE_API_KEY environment variable:

export BRAVE_API_KEY=your-brave-api-key

You can also set the BRAVE_API_KEY in the mcp-server-config.json file. You can get the free BRAVE_API_KEY from Brave Search API.

5. Running from the CLI:

   uv run cli.py
   

   To explore the available commands, use the help option. You can chat with LLM using chat command. Sample prompts:

     What is the capital city of India?
   

     Search the most expensive product from database and find more details about it from amazon?
   

6. Running from the REST API:

   uvicorn app:app --reload
   

   You can use the following curl command to chat with llm:

   curl -X POST -H "Content-Type: application/json" -d '{"message": "list all the products from my local database?"}' http://localhost:8000/chat
   

   You can use the following curl command to chat with llm with streaming:

   curl -X POST -H "Content-Type: application/json" -d '{"message": "list all the products from my local database?", "streaming": true}' http://localhost:8000/chat
   

Contributing

Feel free to submit issues and pull requests for improvements or bug fixes.