gai-commit 0.1.2

An AI-powered git commit message generator.

gai-commit

AI-powered CLI tool to generate commit messages from your staged Git changes. Works with both local Ollama models and OpenAI's API.

Installation

pip install gai-commit

Usage Options

1. Using Ollama (Local LLMs)

Ollama provides free, locally-running AI models with no API keys required.

Prerequisites:

# 1. Install Ollama (https://ollama.com)
# 2. Pull a model (only needed once):
ollama pull llama3.2
# Ollama daemon should start automatically after installation

Basic Usage:

# Stage your changes first
git add path/to/modified/files

# Generate commit message with default model
gai

# Or specify a different Ollama model
gai deepseek-r1:8b

2. Using OpenAI (Cloud-based LLMs)

For higher quality results, you can use OpenAI's models (requires API key).

Prerequisites:

# Set your OpenAI API key (or add to .env file)
export OPENAI_API_KEY=sk-your-key

Basic Usage:

# Stage your changes first
git add path/to/modified/files

# Generate commit message with default model (gpt-3.5-turbo)
gai --provider openai

# Or specify a different OpenAI model
gai --provider openai gpt-4o

Additional Options

# Generate a concise one-line commit message (subject only)
gai --oneline

# Combine with provider selection
gai --provider openai --oneline

How It Works

1. Repository Check: Verifies you're in a git repository via gai.utils.is_git_repository
2. Diff Collection: Gets staged changes with git diff --staged --minimal --unified=5 using gai.utils.get_staged_diff
3. AI Processing: Sends the cleaned diff to the selected AI provider through gai.cli.generate_commit_message
4. Output Cleaning: Formats the AI output and removes technical artifacts via gai.utils.clean_commit_message
5. Interactive Workflow: Presents options to apply, edit, regenerate or quit via gai.cli.handle_user_choice
6. Commit: Applies your approved message with git commit

Interactive Workflow

After generating a commit message suggestion, you'll see:

Suggested Commit Message:
feat(parser): improve error resilience

- add fallback recovery for malformed input
- reduce panic cases in edge parsing paths
---
[A]pply, [E]dit, [R]-generate, or [Q]uit? (a/e/r/q)

Your options:

• A: Apply immediately (git commit -m "<message>")
• E: Open your $EDITOR (defaults to vim) to refine the message
• R: Ask the AI to generate a new suggestion using the same diff
• Q: Quit without committing

Troubleshooting

Common Issues

• "No staged changes found": Use git add to stage your changes first
• "Not a Git repository": Make sure you're inside a valid git repository
• "OPENAI_API_KEY environment variable not set": Set your OpenAI API key or use Ollama
• "Ollama connection refused": Make sure the Ollama daemon is running (ollama serve)

Development

# Clone the repository
git clone https://github.com/muzahid59/gai
cd gai

# Install in development mode
pip install -e .

# Run tests
pytest tests -v

Benchmarking

Compare different models' performance:

# Make sure to stage some changes first
python run_benchmark.py

License

MIT - see LICENSE