ag-llm

A Python package that integrates LLM functionality directly into your Python shell experience.

Installation

uv install ag-llm

Or alternatively:

pip install ag-llm

Usage

Once installed, you can use ag directly in your Python shell with a simple syntax:

import ag
ag/"how are you"

This sends your query to the default LLM and streams the response directly in your terminal with beautiful markdown rendering thanks to Streamdown.

Features

• Direct Python Integration: Use natural Python syntax with the / operator
• Streaming Responses: Get real-time output as the model generates responses
• Markdown Rendering: Beautifully formatted output with code highlighting
• IPython Integration: Automatically hooks into IPython for seamless usage
• Simple Setup: Just import and start querying

Requirements

• Python >= 3.8
• llm for LLM integration
• streamdown for markdown rendering

Configuration

The package uses the default configuration from the llm package for model selection. You can configure your preferred model using:

llm models default your-model-name

See the LLM documentation for more configuration options.

Example

import ag

# Simple query
ag/"What is the capital of France?"

# More complex query
ag/"Explain quantum computing in simple terms"

How It Works

When you use ag/"your query", the package:

1. Sends your query to the configured LLM model via the llm package
2. Streams the response token by token
3. Renders the output using Streamdown for beautiful formatting
4. Handles all the complexity behind the scenes

Development

To develop on this package:

git clone [repository-url]
cd ag-llm
uv install

License

MIT