Continue Commit Notes

AI-powered git commit message generator for VS Code, powered by your locally configured Continue.dev models.

Overview

Continue Commit Notes adds a single-click C button to the Source Control panel in VS Code. Click it and the extension:

1. Reads your staged diff (or all working-tree changes if nothing is staged)
2. Looks up which chat models you have configured in ~/.continue/config.json (or config.yaml) and selects one based on your settings
3. Sends the diff to that model with a structured prompt
4. Writes the generated message directly into the SCM commit input box — ready to review, edit, and commit

No cloud service, no API keys required for local models. Ollama, LM Studio, and llama.cpp work out of the box.

Requirements

The extension reads your Continue configuration (~/.continue/config.json or config.yaml) to discover available models. You do not need the Continue extension to be actively running — as long as the config file is present the extension can call model APIs directly.

Installation

1. Install from the VS Code Marketplace: search "Continue Commit Notes"
2. Make sure Continue.dev is installed and you have at least one model configured
3. Open a git repository — the C icon appears automatically in the Source Control panel

Usage

Generating a commit message

There are three ways to trigger generation:

① Inline button in the Changes / Staged Changes header
Hover over the Changes or Staged Changes group label in the Source Control panel. A C icon appears inline — click it.

② Title bar button
A C icon is always visible in the top-right of the Source Control panel title bar.

③ Command Palette
Open the Command Palette (Ctrl+Shift+P / Cmd+Shift+P) and run:

• Continue Commit: Generate Commit Message (via Continue.dev config)
• Continue Commit: Generate Commit Message (via Continue.dev config) (Pick Model)

What gets sent to the model

• If you have staged changes, only those are included in the diff (git diff --cached)
• If nothing is staged, the full working-tree diff is used, with a warning
• Very large diffs are automatically truncated to 8 000 characters (configurable) to stay within model context limits

Model selection

By default the extension uses the first model listed in your Continue config. You can override this in three ways:

1. Pin a model — set continueCommit.preferredModel to the model's title (or name) field
2. Always show a picker — set continueCommit.showModelPicker: true
3. Per-generation picker — run Continue Commit: Generate Commit Message (via Continue.dev config) (Pick Model) from the palette or right-click menu

Configuration

All settings are under continueCommit.* in VS Code settings (Ctrl+,).

Example settings.json

{
  // Always use Ollama's Llama 3 model, no picker
  "continueCommit.preferredModel": "Ollama Llama 3",
  "continueCommit.showModelPicker": false,

  // Use Conventional Commits format
  "continueCommit.commitStyle": "conventional",

  // Allow longer messages with more context
  "continueCommit.maxTokens": 512
}

How Continue model discovery works

The extension reads ~/.continue/config.json (or config.yaml for newer Continue versions) and looks at the top-level models array. Any entry with both a title and a model field is considered a chat model.

// ~/.continue/config.json (example)
{
  "models": [
    {
      "title": "Ollama Llama 3",   // ← used as the display name and selector key
      "provider": "ollama",
      "model": "llama3"
    },
    {
      "title": "GPT-4o",
      "provider": "openai",
      "model": "gpt-4o",
      "apiKey": "sk-..."
    },
    {
      "name": "qwen2.5-coder-7b-instruct-q4",
      "provider": "llama.cpp",
      "model": "qwen2.5-coder-7b-instruct-q4_k_m.gguf",
      "apiBase": "http://localhost:8000",
      "roles": [
        "apply",
        "chat",
        "edit",
        "autocomplete"
      ]
    }
  ]
}

Supported providers

When useContinueProxy is true (default), the extension first tries to POST to Continue's local proxy at localhost:65432/chat/completions. This means any provider Continue supports will work transparently. If the proxy is not running, the extension falls back to calling the provider API directly using the credentials stored in your config.

Output channel

All activity is logged to the Continue Commit Notes output channel (View → Output → Continue Commit Notes). Check there first if something isn't working.

Commit message formats

Conventional Commits (default)

feat(auth): add refresh token rotation

Tokens are now rotated on every use. The old token is invalidated
immediately, which closes a session-fixation window.

Closes [#127](https://github.com/bishopsmove/continue-based-commit-notes/issues/127)

Types used: feat, fix, docs, style, refactor, perf, test, build, ci, chore, revert

Freeform

Rotate refresh tokens on every use

Eliminates the session-fixation window that existed when a stolen
token remained valid until expiry.

Development

Prerequisites

node >= 18
npm >= 9

Setup

git clone https://github.com/your-username/continue-commit-notes
cd continue-commit-notes
npm install

Build

npm run compile          # development build → dist/extension.js
npm run watch            # rebuild on save
npm run vscode:prepublish  # minified production build

Run tests

npm test
# Compiles TypeScript to out/, then launches @vscode/test-electron
# which downloads a headless VS Code and runs the Mocha test suite.

Package for Marketplace

vsce package          # produces continue-commit-notes-x.y.z.vsix

Install the VSIX locally with Extensions: Install from VSIX… in VS Code, or publish with vsce publish.

Contributing

Pull requests are welcome. For significant changes please open an issue first to discuss the approach.

1. Fork the repository
2. Create a feature branch: git checkout -b feat/my-feature
3. Make your changes and add tests
4. Verify: npm test
5. Open a pull request

License

MIT — see LICENSE for details.