Automatic commit message generator using AI, with some extra features. But first, why?
Writing commit messages is often a tedious task, and it is not always easy to condense or describe the changes we have made (or the bugs we have introduced), but fortunately this message is not as critical as the code we develop. AI can do a lot of things, although the less important that task is, the better. So for this one, it could help us generate meaningful and descriptive messages automatically, adding those “best practices” that, of course, we all apply in our commit messages. And yes, it creates the title and the description.
- 🧠 Multiple AI providers: OpenAI, Google Gemini, Azure OpenAI and Ollama
- 🌍 Multi-language support: Generate commits in Spanish, English, French, German, etc.
- 📋 Interactive staging: Select which files to include in the commit
- ✅ User validation: Ask for user validation before committing
- ➕ Additional context: Use extra information to generate more accurate messages, like branch name or manual context
You can install AutoCommit AI directly from PyPI using pip:
pip install auto-commit-ai
Alternatively, you can clone the repository and install directly from source:
git clone https://gitlab.com/tmllull/auto-commit-ai.git
cd auto-commit-ai
pip install .
AutoCommit AI loads its configuration from environment variables file. You can set these variables in a .auto_commit_ai.env
file located in your current working directory (where your git repository is) or your home directory (~/.auto_commit_ai.env).
NOTE: If you add the file to your working directory, remember to add it to your .gitignore
file.
The contents of the .auto_commit_ai.env
can be found in the .auto_commit_ai.env.template
file.
AutoCommit AI is a command-line tool. Navigate to your Git repository and run one of the following commands:
# Default command
auto-commit-ai
or
# Alias for the default command
acai
This is the default command, and it will generate a commit message for your staged changes only.
Both commands, auto-commit-ai
and acai
, are equivalent and can be used interchangeably.
# Generates a commit message only for staged changes
auto-commit-ai
# Show help message
auto-commit-ai -h
# Include all changes (staged + unstaged + untracked)
acai --all
# Use specific provider
acai --provider google
# Generate in Spanish
acai --language es
# Preview without committing option
auto-commit-ai --preview
# Preview including all changes
acai --preview --all
# Select files for staging
acai --stage
# Repository status
acai --status
# Commit history
auto-commit-ai --history
# Specific repository
auto-commit-ai --repo /path/to/repo
# Specific repository
auto-commit-ai --custom-prompts /path/to/custom_prompts.py
# Use current branch name in commit context
acai --branch-name
# Additional context for commit message generation
acai --context "Some extra context here"
--custom-prompts
flag: Use this flag by your own risk. It allows you to use custom prompts for commit message generation, but it may lead to unexpected results if the prompts are not properly formatted or tested. Check the custom_prompts_example.py
file for an example of how to create your own prompts.
# Don't show status before commit
acai --no-status
# Verbose mode for debugging
acai --verbose
The system supports ISO 639-1 language codes:
en
- English (default)es
- Spanishfr
- Frenchde
- Germanit
- Italianpt
- Portugueseja
- Japanesezh
- Chineseru
- Russian- And many more... As you use the ISO 639-1 codes, the system will adapt to your needs.
# 1. Check repository status
acai --status
# 2. Interactive file staging
acai --stage
# 3. Preview message
acai --preview
# 4. Generate final commit
acai
# Work with another repository
acai --repo /path/to/project --status
acai --repo /path/to/project --all
- Analysis: Detects repository changes (staged, unstaged, untracked)
- Generation: Uses AI to analyze diffs and generate descriptive message
- Review: Shows generated message for confirmation
- Commit: Creates commit with generated message
- Optional push: Offers to push automatically
Generated messages follow best practices:
- Title: Concise line describing the main change
- Description: Additional details when necessary
- Conventions: Follows standards like Conventional Commits when appropriate
The system gracefully handles:
- ❌ Non-Git repositories
- ❌ No changes to commit
- ❌ AI API errors
- ❌ Connectivity issues
- ❌ Incorrect configuration
This project is licensed under the MIT License. See LICENSE
file for more details.
To report bugs or request features, please open an issue on the GitHub repository.
Made with ❤️ for developers who value quality commits