CommitMate Local

Generate concise Git commit messages inside VS Code — powered entirely by local Ollama models.

No cloud API. No subscription. No data leaving your machine.

⚠️ Beta. This extension is under active development. If something breaks, please open an issue.

What it does

CommitMate Local reads your staged Git diff and asks a local Ollama model to write a concise, Conventional Commit-style message — right from the Source Control panel. Everything runs on your machine, so your code diffs never leave localhost.

Features

• ✅ Generate commit messages directly from the Source Control view
• ✅ Analyzes only staged changes (git diff --cached)
• ✅ Runs fully locally through Ollama — no external AI API required
• ✅ Ships with qwen2.5-coder:3b as the default model
• ✅ Produces Conventional Commit-style messages (feat:, fix:, chore:, …)
• ✅ Auto-fills the commit input box — review and edit before committing
• ✅ Skips generation entirely when nothing is staged

Requirements

Before using this extension, make sure you have:

• Ollama installed and running locally

• The default model pulled:

  ollama pull qwen2.5-coder:3b
  

Start the Ollama server before generating commit messages:

ollama serve

By default, Ollama listens on http://127.0.0.1:11434.

Getting Started

1. Install and start Ollama, then pull the default model (see Requirements).
2. Open a Git repository in VS Code.
3. Stage the changes you want to commit (git add <file>, or use the Source Control view).
4. Open the Source Control panel.
5. Click Generate Local Commit Message.
6. Review the generated message in the commit input box — edit if needed.
7. Commit as usual.

Example output:

feat: add membership cancellation eligibility validation

Extension Settings

(Update this table to match your actual contributes.configuration entries in package.json.)

Privacy

CommitMate Local is designed to work exclusively with a locally running Ollama server. Staged Git diff content is sent only to your local endpoint (http://127.0.0.1:11434 by default) for commit-message generation. No external AI API key is required, and no diff content leaves your machine under the default configuration.

Known Issues

• No staged changes → nothing to generate. Stage at least one change first.
• If Ollama isn't running, generation will fail — start it with ollama serve.
• If the configured model isn't installed, pull it with ollama pull qwen2.5-coder:3b.

See the issue tracker for the latest known issues.

Release Notes

See CHANGELOG.md for details on each release.

0.1.0

Initial beta release — staged-diff commit message generation via local Ollama models.

Feedback & Contributing

Bug reports, feature requests, and pull requests are welcome on GitHub.

License

MIT