Mcpx Py

🚀 mcpx-py Project

A Python library for interacting with LLMs via the mcp.run tool.

✨ Features

Supported AI Providers

mcpx-py supports all models supported by PydanticAI.

📦 Installation Guide

Dependencies

• uv
• npm
• ollama (Optional)

mcp.run Configuration

You need to obtain an mcp.run session ID by running the following command:

npx --yes -p @dylibso/mcpx gen-session --write

This will generate a new session and write the session ID to the configuration file, which can be used by mcpx-py.

If you need to store the session ID in an environment variable, you can run gen-session without the --write flag:

npx --yes -p @dylibso/mcpx gen-session

This should output something like:

Login successful!
Session: kabA7w6qH58H7kKOQ5su4v3bX_CeFn4k.Y4l/s/9dQwkjv9r8t/xZFjsn2fkLzf+tkve89P1vKhQ

Then set the MPC_RUN_SESSION_ID environment variable:

$ export MPC_RUN_SESSION_ID=kabA7w6qH58H7kKOQ5su4v3bX_CeFn4k.Y4l/s/9dQwkjv9r8t/xZFjsn2fkLzf+tkve89P1vKhQ

Installation Methods

Python Usage Instructions

Example Code

Here is a complete example showing how to use mcpx-py:

from mcpx import MPX

# Initialize an MPX instance
mpx = MPX()

# Perform inference using different LLM models
print(mpx.chat("What is the weather like?"))  # Use the default model
print(mpx.chat("What is the weather like?", model="gpt-3.5-turbo"))  # Use a specified model

# Process structured output
response = mpx.chat("Tell me a joke.", result_type=dict)
print(response)

Install Dependencies

Before using mcpx-py, make sure to install all the necessary dependencies:

pip install mcpx

Command-Line Usage Instructions

Tool Installation

Install using the uv Package Manager

Add mcpx to your uv configuration file (usually located at ~/.config/uv.json):

{
  "dependencies": {
    "mcpx": "*"
  }
}

Then run the following command to install:

uv install

Install using the npm Package Manager

Although it is mainly installed via pip, you can also use npm for installation (although this is not the recommended method):

npm install mcpx-py

Basic Usage

Chat with the LLM

Run the following command to start chatting with the LLM:

mcpx-client chat

Get Help Information

If you need more information, you can run:

mcpx-client --help

📚 Documentation

LLM Configuration Guide

Supported Models and Their Configuration Methods

Use the Gemini Model

1. Create an account on Gemini.
2. Generate an API key and follow the instructions for use.

Use Ollama

1. Download and install Ollama: ollama.ai
2. Pull the required model, for example:

ollama pull llama3.2

3. Set it as the default model:

ollama config default=llama3.2

Use Llamafile

1. Download a Llamafile model: llamafile
2. Set the file as executable:

chmod +x your-model.llamafile

3. Run it in JSON API mode:

./your-model.llamafile --json-api --host 127.0.0.1 --port 8080

4. Use the OpenAI provider to point to the local service address: http://localhost:8080

List of Supported Models

• gpt-3.5-turbo
• llama2
• llama3
• llama3.2

Set the API Key

Add your API key to the environment variable, for example:

export OPENAI_API_KEY="your-api-key"

In this way, you can interact with these models.

Through the above steps, you can easily configure and use various LLM models to meet different development needs.