
Get started with Gyti in two simple steps:
1. Install globally:
2. Run setup (configures everything automatically):
✓ That's it! You're ready to go.
The setup command will configure Gyti, authenticate you, and install the Git hook automatically. You can now use git commit as usual and Gyti will generate your commit messages.
gyti setup do?The setup command guides you through:
On first run, Gyti will automatically prompt you to run setup if you haven't configured it yet.
If you're developing or modifying the CLI, you need to rebuild after making changes:
This rebuilds the TypeScript code and updates the globally linked command.
Once installed, just commit as usual. The tool will automatically generate or enhance your commit message:
You can also use git commit -m "." and the tool will replace it with a generated message.
Test the message generation without committing:
Check your current settings:
Set the backend API URL (default: https://www.gyti.ai/api):
Set your API key for authentication (if required by your backend):
Your API key is stored securely in your system's config directory.
Enable or disable verbose logging:
Skip AI generation globally (useful for testing or when you want to write messages manually):
Enable or disable interactive refinement of commit messages (enabled by default):
When using the Gyti backend, you need to authenticate with your account:
Authenticate with email/password or OAuth (GitHub, Google):
This will open your browser for authentication. Press Enter when prompted to open the browser.
Sign out and clear authentication tokens:
You can also configure Gyti using environment variables:
GYTI_SKIP=trueSkip AI generation for a specific commit. Useful for one-off commits.
GYTI_VERBOSE=trueEnable verbose output for debugging. Shows detailed error messages and generation status.
GIT_GYTI_SKIP=trueAlternative environment variable name for skipping AI generation.
Gyti requires a backend API to generate commit messages. You can use the provided Next.js backend or set up your own.
Start the Next.js backend server:
The backend will run on http://localhost:3000 by default.
You can use your own AI provider instead of the Gyti backend. Run setup and choose "Self-hosted":
Supported providers:
Make sure you ran gyti install in your repository.
Check your backend URL and ensure the server is running:
The tool automatically retries failed requests. If timeouts persist:
GYTI_VERBOSE=true git commitTo remove Gyti from a repository: