Nuvo Commit

Write commits, not novels. Local-first AI commit message generator.

Watch in full quality

Features

• 🤖 Multi-Provider Support: Choose from Ollama (local), OpenAI, Anthropic, or the models already in VS Code (e.g. GitHub Copilot)
• 🆓 No Key Needed with Copilot: The VS Code provider reuses your Copilot models — no API key, no local server
• 🔒 Local-First: Run entirely on your machine via Ollama — no cloud APIs required
• ☁️ Cloud Options: Use OpenAI or Anthropic Claude for higher-quality results
• 🔑 Secure Keys: Cloud API keys are stored in VS Code's encrypted secret storage
• 🌍 Any Language: Write commit messages and PR content in your team's language
• 🎯 Matches Your Repo: Learns the style of your recent commits and follows the allowed types and scopes in your commitlint config
• 🚀 Fast: Quick commit message generation without leaving VS Code
• ⚙️ Customizable: Configure models, endpoints, timeouts, and behavior
• 📋 Auto Model Detection: Automatically list available models from your provider

Requirements

• VS Code 1.90.0 or higher
• For Ollama (local): Ollama installed and running, with a model pulled (ollama pull qwen3:4b)
• For OpenAI (cloud): an API key from https://platform.openai.com/api-keys
• For Anthropic (cloud): an API key from https://console.anthropic.com/
• For VS Code (Copilot): GitHub Copilot (or another extension that provides language models) installed and signed in. VS Code asks for permission the first time Nuvo Commit uses a model.

Installation

1. Install the extension from the VS Code Marketplace.
2. Choose your provider:
   • Local (recommended): install Ollama (brew install ollama on macOS, or see ollama.ai).
   • Cloud: get an API key from OpenAI or Anthropic.
3. Follow the Get Started with Nuvo Commit walkthrough that opens after install (or configure manually, see Configuration).
4. Run Nuvo Commit: Check Setup to confirm the provider and model work. If something is wrong, the error offers the fix.
5. Start generating commit messages!

Usage

1. Stage your changes (git add .). If nothing is staged, Nuvo Commit falls back to your working-directory changes.
2. Open the Command Palette (Cmd+Shift+P / Ctrl+Shift+P).
3. Run Nuvo Commit: Generate Commit Message.
4. Review and accept the suggested message.

You can also click the ✨ button in the Source Control title bar.

Tip: With the Source Control view focused, press Cmd+Alt+G / Ctrl+Alt+G to generate. Change it under Keyboard Shortcuts.

Generating pull request content

1. Click the pull request button in the Source Control title bar, or run Nuvo Commit: Generate Pull Request Content.
2. Pick the target branch (the default is listed first).
3. Accept to copy the title and body, copy them separately, or open GitHub's "create pull request" page.

Selecting a model

1. Run Nuvo Commit: Select Model.
2. Pick from the discovered models, or choose "Enter manually…".
3. The selection is saved to your settings.

Tip: For Ollama, the model list is fetched automatically from your local instance.

Setting an API key (cloud providers)

API keys are stored securely per cloud provider — not in settings.json.

1. Run Nuvo Commit: Set API Key.
2. If Ollama is active, choose OpenAI or Anthropic.
3. Paste your key into the masked input (leave it empty to clear that provider's key).

If you previously set nuvoCommit.apiKey in settings, it is migrated into secure storage automatically and removed from settings.json on first activation.

Configuration

Run Nuvo Commit: Settings to configure only the selected provider's model, endpoint, and API key. Provider-specific settings are also available in VS Code Settings (Cmd+,) under Nuvo Commit:

Deprecated fallback settings nuvoCommit.apiKey, nuvoCommit.endpoint, and nuvoCommit.model are kept for upgrades. New configurations should use the provider-specific settings above.

Example configurations

Ollama (local):

{
  "nuvoCommit.provider": "ollama",
  "nuvoCommit.ollama.endpoint": "http://localhost:11434/api/generate",
  "nuvoCommit.ollama.model": "qwen3:4b"
}

OpenAI (cloud): set the key via Nuvo Commit: Set API Key, then:

{
  "nuvoCommit.provider": "openai",
  "nuvoCommit.openai.endpoint": "",
  "nuvoCommit.openai.model": "gpt-4o-mini"
}

Anthropic (cloud): set the key via Nuvo Commit: Set API Key, then:

{
  "nuvoCommit.provider": "anthropic",
  "nuvoCommit.anthropic.endpoint": "",
  "nuvoCommit.anthropic.model": "claude-haiku-4-5"
}

VS Code / GitHub Copilot (no API key):

{
  "nuvoCommit.provider": "vscode",
  "nuvoCommit.vscode.model": ""
}

OpenAI-compatible providers

The openai provider works with any service that speaks the OpenAI Chat Completions API. Set nuvoCommit.provider to openai, point nuvoCommit.openai.endpoint at the service's /chat/completions URL, and save the service's key with Nuvo Commit: Set API Key (choose OpenAI). A key is always required; for local servers that don't check it, enter any value. Nuvo Commit: Select Model lists the models the service offers.

Google Gemini (API key):

{
  "nuvoCommit.provider": "openai",
  "nuvoCommit.openai.endpoint": "https://generativelanguage.googleapis.com/v1beta/openai/chat/completions",
  "nuvoCommit.openai.model": "gemini-3.5-flash-lite"
}

OpenRouter (API key):

{
  "nuvoCommit.provider": "openai",
  "nuvoCommit.openai.endpoint": "https://openrouter.ai/api/v1/chat/completions",
  "nuvoCommit.openai.model": "openai/gpt-4o-mini"
}

Groq (API key):

{
  "nuvoCommit.provider": "openai",
  "nuvoCommit.openai.endpoint": "https://api.groq.com/openai/v1/chat/completions",
  "nuvoCommit.openai.model": "llama-3.3-70b-versatile"
}

LM Studio (local; start the server in LM Studio's Developer tab):

{
  "nuvoCommit.provider": "openai",
  "nuvoCommit.openai.endpoint": "http://localhost:1234/v1/chat/completions",
  "nuvoCommit.openai.model": "<model id shown in LM Studio>"
}

Tip: Prefer non-reasoning models. Reasoning models (e.g. gpt-oss, qwen3) can spend the whole response budget on hidden reasoning and return nothing, in which case you get the generic chore: update staged changes message.

Development

npm install      # Install dependencies
npm run compile  # Compile TypeScript
npm run watch    # Compile in watch mode
npm run lint     # Lint sources
npm run format   # Format with Prettier
npm run test:unit       # Run Jest unit tests
npm run test:coverage   # Run unit tests with coverage

Debugging

1. Open the project in VS Code.
2. Press F5 to launch the extension in a new Extension Development Host window.

Testing

See TESTING.md for detailed testing instructions.

Publishing

See PUBLISHING.md for the release and publishing guide.

Changelog

See CHANGELOG.md.

License

MIT