@chris_glolabs/git-ai-commit NPM

Git AI Commit

Generate concise Git commit messages using the power of local AI models with Ollama.

git-ai-commit analyzes your staged changes and uses a local large language model (LLM) via Ollama to suggest a relevant and brief commit message. Streamline your workflow and let AI handle the first draft of your commit messages!

Installation

Install the package globally using npm:

npm install -g @chris_glolabs/git-ai-commit

The installation process automatically runs a setup script that:

Checks if Git is installed.
Sets up a global Git alias: git aic which executes git-ai-commit.

You can now use git aic in any Git repository on your system.

Features

✨ AI-Powered Commits: Leverages local LLMs through Ollama for message generation.
🏡 Local First: Keep your code private. No data is sent to external servers (besides your local Ollama instance).
🚀 Simple Workflow: Integrates seamlessly with your Git process via a handy alias (git aic).
⚙️ Configurable: Choose your preferred Ollama model, set timeouts, customize prompts, and decide whether to auto-accept suggestions.
🤖 Automatic Setup: Installs a convenient git aic alias during setup.
✅ Interactive Confirmation: Review and confirm the generated message before committing (configurable).

Prerequisites

Before you begin, ensure you have the following installed:

Git: Download Git
Node.js: (Version 16 or higher recommended) Download Node.js
Ollama: Needs to be installed and running.
- Download Ollama
- Important: Make sure the Ollama server is running before using git-ai-commit. You can typically start it by running ollama serve in your terminal.
- Ensure you have pulled at least one model (e.g., ollama pull llama3:8b).

Usage

Using git-ai-commit is straightforward:

Stage your changes: Make your code changes and stage them using git add.

# Example: Stage all changes
git add .

# Example: Stage specific files
git add src/feature.js src/utils.js

Generate the commit message: Run the git aic command.
```
git aic
```
Review and Confirm:
- The tool will connect to your running Ollama instance, analyze the staged diff, and generate a commit message.
- It will display the suggested message and ask for your confirmation.
- Press Enter (or y then Enter) to accept and commit.
- Press n then Enter to abort the commit.
(You can disable the confirmation step via configuration - see below)

Alternative Commands:

You can also invoke the tool directly:

# Equivalent to 'git aic'
git-ai-commit

# Explicitly use the 'commit' command (same as default)
git-ai-commit commit

Configuration

You can customize git-ai-commit's behavior using the config command:

git aic config
# or directly via
git-ai-commit config

This will launch an interactive prompt allowing you to configure:

Model Selection: Choose which Ollama model to use (e.g., llama3:8b, mistral). It attempts to list models available locally if Ollama is running. Ideally choosing a model that will run comfortably on your machine.
Always Accept: Set to true to skip the confirmation step and automatically use the generated commit message. Defaults to false.
Timeout: Set the timeout (in seconds) for requests to the Ollama API. Increase this if you use larger models or have slower hardware. Defaults to 30 seconds.
Custom Prompt: Allows you to provide your own prompt template for generating commit messages. You can edit the prompt directly in your default text editor.

Your configuration is saved in ~/.git-ai-commit/config.json.

Troubleshooting

Error: Ollama is not installed: Make sure you have installed Ollama correctly. See Prerequisites.
Error: Ollama is not running: Start the Ollama server, usually by running ollama serve in your terminal.
Error: Request timed out / Ollama is taking too long:
- Your selected model might be too large for your system or the changes are too complex.
- Try configuring a smaller model (e.g., llama3:8b instead of llama3:70b) using git-ai-commit config.
- Increase the timeout using git-ai-commit config.
Error: Failed to fetch available models: This might happen if Ollama isn't running or accessible when you run git-ai-commit config. You can still manually type the name of a model you know you have installed.