Commit Assistant (VS Code Extension)
Generate Git commit messages with AI models — powered by GitHub Copilot or your own custom AI endpoints.
Features
- AI-Powered Commit Messages — Generate commit messages from git diff and related history, or purely from your own description.
- Optional Intent Input — Describe what you intend to commit; leave empty to auto-analyze changes. When provided, the message is generated solely from your description.
- Multiple AI Models — Use GitHub Copilot models or add custom OpenAI-compatible endpoints.
- Multiple Languages — Generate commit messages in 15 supported languages.
- Commit Styles — Choose from conventional, angular, semantic, emoji, and more.
- Source Control Integration — One-click button in the SCM panel to generate messages.
- Multi-Root Workspace — When multiple git repositories are open, select the target repository from a quick pick.
- Usage Statistics — Track daily/monthly generation counts and token usage per model via the status bar and dashboard.
- Bilingual UI — Switch the extension UI between English and Chinese. All prompts, messages, validations, and errors are fully localized.
Commands
- Commit Assistant: Generate Commit Message
- Prompts for optional user intent. Leave empty to auto-analyze changes.
- The generated message is inserted into the SCM input box (when available) and copied to clipboard.
- Commit Assistant: Switch Model
- Switch between GitHub Copilot models or custom models.
- Format:
copilot:<model> or custom:<name>.
- Commit Assistant: Switch Style
- Select commit message style:
auto, plain, conventional, angular, karma, semantic, emoji, emojiKarma, google, atom.
- Commit Assistant: Switch Language
- Select the output language for generated commit messages.
- Commit Assistant: Switch UI Language
- Switch command titles and messages between English (
en) and Chinese (zh).
- Commit Assistant: Add Custom Model
- Register a custom AI model by providing name, API key, URL, and model ID.
- API keys are securely stored in VSCode SecretStorage.
- Commit Assistant: Remove Custom Model
- Remove a previously added custom model.
- Commit Assistant: List Custom Models
- View all configured custom models.
- Commit Assistant: Show Statistics
- Open a dashboard showing daily/monthly usage and token consumption per model.
- Commit Assistant: Clear Statistics
- Reset all usage statistics.
Settings
| Setting |
Type |
Default |
Description |
commitAssistant.model |
string |
copilot:gpt-4.1 |
Preferred model. Use copilot:<model> for GitHub Copilot or custom:<name> for custom models. |
commitAssistant.style |
enum |
conventional |
Commit message style mode. |
commitAssistant.language |
enum |
english |
Output language for commit messages. |
commitAssistant.uiLanguage |
enum |
en |
UI language (en or zh). |
commitAssistant.maxDiffChars |
number |
16000 |
Maximum diff characters sent to the model (2000–60000). |
commitAssistant.stagedOnly |
boolean |
false |
Only analyze staged changes when generating commit messages. Unstaged and untracked changes are ignored. |
commitAssistant.customModels |
array |
[] |
Custom model configurations (name, url, model). |
Supported Output Languages
english, chinese, spanish, french, german, japanese, korean, russian, portuguese, italian, dutch, turkish, polish, vietnamese, arabic
Custom Models
You can use any OpenAI-compatible API endpoint as a custom model:
- Run Commit Assistant: Add Custom Model.
- Enter a unique name (e.g.,
my-openai).
- Enter your API key — it will be stored securely in VSCode SecretStorage.
- Enter the API endpoint URL (e.g.,
https://api.openai.com/v1/chat/completions).
- Enter the model name (e.g.,
gpt-4o).
Then set commitAssistant.model to custom:my-openai.
Usage Statistics
A status bar item appears at the bottom-right of VS Code. Hover to see a formatted table with today's and this month's generation counts per model. Click to open a detailed dashboard with:
- Total generations per model
- Token usage per model
- Daily and monthly usage breakdowns with bar charts
Statistics older than 180 days are automatically pruned.
Development
npm install
npm run esbuild # bundle with esbuild
npm run esbuild:watch # watch mode
npm run build # compile with tsc (type checking)
npm test # run tests
Run extension in debug host:
- Open this folder in VS Code.
- Press
F5 to launch Extension Development Host.
- Open Command Palette and run one of the commands above.
Notes
- GitHub Copilot models require the GitHub Copilot extension and a valid subscription.
- Custom models use OpenAI-compatible chat completions format.
- URL security validation blocks private IPs and localhost for safety.
- Token usage for Copilot models is estimated (~4 chars per token); custom models report actual usage from the API response.
| |
