What is Elixir Mcp

Elixir MCP is a modular implementation of the Model Context Protocol (MCP) designed for Elixir. It facilitates communication between machine learning models and applications by providing a standardized framework for understanding and processing contextual information.

Use cases

Use cases for Elixir MCP include building intelligent chatbots that understand user context, developing applications that require contextual data for machine learning models, and creating servers that facilitate communication between different AI tools and models.

How to use

To use Elixir MCP, you can create an MCP server by defining a module in Elixir. The server will handle requests and responses, enabling LLMs to call tools and access context effectively.

Key features

Key features of Elixir MCP include modular design, support for multiple protocols, ease of integration with existing Elixir applications, and the ability to enhance the performance of NLP models through improved context handling.

Where to use

Elixir MCP can be used in various fields such as natural language processing (NLP), machine learning applications, AI-driven chatbots, and any system requiring contextual understanding for better predictions.

Model Context Protocol SDK for Elixir

A toolkit for building servers (and clients!) that enable LLMs to call tools and access context.

Note: This project is currently a work in progress. Features and APIs may change as development continues.

Overview

Elixir MCP is a modular Elixir project that provides a protocol implementation and server for the MCP (Model Context Protocol). The Model Context Protocol is a framework that defines how machine learning models, particularly those used in NLP, understand and process contextual information. It establishes standardized methods for providing models with appropriate context and background data to improve their understanding and responses. MCP enables consistent communication between AI models and applications, allowing for more accurate and relevant predictions by establishing rules for how models take in, process, and respond to contextual information in a distributed environment.

For more information about the Model Context Protocol, visit the official documentation.

Getting started - creating an MCP server

defmodule MyServer do
   use MCP.Server, name: "My cool server", version: "1.0.2"

   @impl true
   def init(_args) do
      state = %{}
      {:ok, state}
   end

   @type location_info :: %{
      place_name: String.t()
   }

   @doc """
   Retrieve information about the given document
   """
   @decorate tool_call()
   @spec get_location_info(input :: String.t(), state ::term()) :: location_info()
   def get_location_info(input) do
     %{place_name: input <> "'s place"}
   end
end

All public functions that are decorated with :tool_call in this module are exposed on the MCP server as tool calls.

Components

MCP Protocol (mcp_protocol/)

Contains the core protocol implementation for Model Context Protocol, parsing and valiting its JSON messages. The protocol’s structs are generated from the official spec.

MCP Server (mcp_server/)

Server implementation for the Model Context Protocol, with flexible adapter for communcation to switch between using STDIO or Server-sent events (Plug).

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.