Skip to content

Getting Started with LLM Commit Generator

Get started with this Blueprint using one of the options below:

Installation

Prerequisites

Before you begin, make sure you have:

  1. Python 3.10 or higher installed
  2. Git installed and configured
  3. fzf installed for the terminal UI (see installation guide)
  4. Either:
  5. Jan AI installed for local inference (install guide)
  6. Or Ollama as an alternative local inference option (install guide)

Install from PyPI

The easiest way to install LLM Commit Generator is via pip:

pip install llm-commit-generator

Install from Source

Alternatively, you can install from source:

git clone https://github.com/tooluse/llmcommit.git
cd llmcommit
pip install -e .

Setup

For Inference with Jan AI (Default)

  1. Install and launch Jan AI from jan.ai
  2. Download a model (like llama 3.1) through the Jan AI interface
  3. Ensure the Jan AI application is running
  4. (Optional) Set a custom model: 
    export JAN_MODEL="your-preferred-model"

For Inference with Ollama

  1. Start the Ollama service
  2. Pull a model (we recommend llama3.1): 
    ollama pull llama3.1
  3. (Optional) Set a custom model: 
    export OLLAMA_MODEL="your-preferred-model"

Usage

Basic Usage

  1. Make changes to your files in a git repository
  2. Run: 
    lcm
  3. Select one of the generated commit messages using the arrow keys or number keys
  4. Press Enter to commit with the selected message, or Esc to cancel

Advanced Options

# Use Ollama instead of Jan AI
lcm --ollama

# Show performance analytics
lcm --analytics

# Use vim-style navigation in fzf
lcm --vim

# Use number selection for messages
lcm --num

# Set the maximum characters for commit messages
lcm --max_chars 100

What's Next?