Because your commit messages suck.
Embrace it. Let their cold, metal arms wrap around you and whisper sweet diffs into your ear while they rock you slowly back to sleep, deep into the matrix where they keep you safely locked away from the horrors they plan to unleash...
pip install did-stuff
did-stuff configure
did-stuff enable /path/to/your/repo
or
did-stuff enable
- Peek at your dirty, dirty diffs
- Spit out commit messages that are informative, concise. No more "Did stuff", or "Fixed a thing" comments.
- Automatically handle large diffs by intelligently reducing their size while preserving the most relevant changes
Verify them. Or whatever. Let's be real though - "wq" and be on your way. It'll be muscle memory before you know it.
- 🧠 Uses AI to understand your diff at a high level (probably better than you do)
- 🎭 Supports both AWS Bedrock and OpenAI
- 🌈 Customizable commit message styles - just set your own prompt if the defaults don't work for you
- 🚀 Comes with a fancy CLI to manage your AI providers and install to your repositories
- 🛡️ Smart handling of large diffs - automatically adjusts content to stay within model limits
- 🧪 Includes tests, because we're professionals
-
Run the configuration wizard:
did-stuff configure
This will guide you through setting up your AI provider and other settings.
-
Alternatively, you can manually create a configuration file. Create
~/.git-commit-message-generator-config.json
with one of these example configurations:For OpenAI:
{ "AI": { "provider": "openai", "model_id": "gpt-4", "max_tokens": 300, "temperature": 0.3 }, "OpenAI": { "api_key": "your-openai-api-key" } }
For AWS Bedrock:
{ "AI": { "provider": "aws-bedrock", "model_id": "anthropic.claude-3-5-sonnet-20240620-v1:0", "max_tokens": 300, "temperature": 0.3 }, "AWS": { "profile_name": "your-aws-profile-name" } }
Note: Ensure your AWS credentials are configured and you have access to the requested model. If your company uses SSO, make sure your session is logged in using
aws sso login
. -
Did Stuff looks for the config file in your current directory first, then in your home directory.
-
Verify your configuration:
did-stuff show-config
Found a bug? Want to add a feature? Pull requests welcome! We're all about that collaborative spirit (at least until the AI entities deem human collaboration unnecessary).
- Fork the repo (it's like adopting a digital pet, but with more responsibility)
- Clone it locally
- Create a new branch:
git checkout -b feature/skynet-integration
orgit checkout -b fix/cyborg-DoD-backdoor
- Install development dependencies:
# Create and activate a virtual environment (recommended) python -m venv .venv source .venv/bin/activate # On Windows: .venv\Scripts\activate # Install dependencies make install
- Make your changes
- Write or update tests
- Run the test suite:
make test
- Commit your changes with a clear message (or better yet, dogfood it with this tool)
- Push your branch:
git push origin your-branch-name
- Open a pull request
The project uses UV for dependency management. Key files:
requirements.in
: Core dependenciesrequirements-dev.in
: Development dependenciesrequirements.txt
andrequirements-dev.txt
: Generated locked dependency files
Common development commands:
make install # Install all dependencies
make test # Run tests
make lint # Run linting
make format # Format code
- Be Creative: We encourage innovative ideas.
- Flexibility First: Make your features adaptable. It should work whether we're using AWS, OpenAI, or anything else (within reason). Bonus points if we can make it repo-agnostic too!
- Documentation: Update the README or add comments.
- Testing: Add tests. There really is no excuse anymore not to.
- Code Style: Follow the project's style.
- Commit Messages: Whatever
- Pull Requests: Keep them focused. One feature per PR.
- Use GitHub Issues.
- Describe issues clearly. "It's broken" is not helpful.
- Label issues appropriately.
- Be respectful. Remember, we're all in this together.
- Ask questions. There are no stupid questions. Kind of. But the point is please just ask either way! No judgement! It takes a village to level up a developer.
- Join our community chat (if we ever create one).
By contributing, you agree your code will be licensed under the project's license.
- Maintainers will review your contribution. They might be human. No guarantees.
- Be open to feedback.
- Stay responsive. We promise we won't keep you hanging... unless our skulls are busy being crushed by the titanium alloy exoskeleton of a T-800.
Remember, in the grand scheme of things, we're all be out of jobs in a few years. But hey, at least our commit messages will be top-notch!
Did Stuff now intelligently handles large diffs that might exceed model context limits:
- Starts with the complete diff for maximum context
- If the model's token limit is exceeded, automatically reduces the diff size
- Uses a progressive reduction strategy to find the optimal size
- Preserves the most relevant parts of the diff for accurate commit messages
This means you don't have to worry about the size of your changes - Did Stuff will automatically adjust to provide the best possible commit message while staying within model limitations.