CommitCraft
Generate git commit messages with AI. Supports any OpenAI-compatible API including proxies.
简体中文 | English
Features
- 🤖 Generate commit messages from your git diff using AI
- 🌐 Works with any OpenAI-compatible API — OpenAI, DeepSeek, Moonshot, Zhipu, Tongyi Qianwen, or any proxy
- 🎨 Three styles: Conventional Commits, Simple, Emoji
- 📝 Two detail levels: Concise / Detailed
- 🌍 7 languages supported
- 📌 Button in the Source Control sidebar title bar
Requirements
- VSCode 1.85+
- An API key for any OpenAI-compatible service
Getting Started
Open the configuration panel:
- Command Palette:
Ctrl+Shift+P → search CommitCraft: Open Settings
- Or click the ⚙️ button in the Source Control title bar
Select a Provider (OpenAI, DeepSeek, OpenRouter, etc.), enter your API Key, and choose a Model. The API Base URL is auto-filled based on your provider choice.
2. Generate Commit Message
- Stage your changes in Git
- Click the 🪄 button in the Source Control title bar
- The generated commit message appears in the input box
Quick Configuration Examples
| Provider |
API Base URL |
Notes |
| OpenAI |
https://api.openai.com/v1 |
Default, just add your key |
| DeepSeek |
https://api.deepseek.com/v1 |
Select "DeepSeek" in provider dropdown |
| OpenRouter |
https://openrouter.ai/api/v1 |
Access all models with one key |
| Any proxy |
Your proxy URL |
Select "Custom" provider |
Configuration
| Setting |
Description |
Default |
commitCraft.apiBaseUrl |
API base URL (OpenAI-compatible) |
https://api.openai.com/v1 |
commitCraft.apiKey |
API key |
(empty) |
commitCraft.presetModel |
Preset model selection (dropdown) |
gpt-4o-mini |
commitCraft.customModel |
Custom model name (overrides preset) |
(empty) |
commitCraft.language |
Commit message language |
English |
commitCraft.style |
Style: conventional / simple / emoji |
conventional |
commitCraft.detail |
Detail level: concise / detailed |
concise |
commitCraft.maxDiffLength |
Max diff characters sent to the model |
8000 |
Preset Models
| Provider |
Models |
| OpenAI |
gpt-4o, gpt-4o-mini, o3-mini |
| Anthropic |
claude-sonnet-4, claude-opus-4, claude-3-5-haiku (via proxy) |
| Google |
gemini-2.5-flash, gemini-2.5-pro |
| DeepSeek |
deepseek-chat, deepseek-reasoner |
| Moonshot |
moonshot-v1-8k |
| Zhipu |
glm-4-plus |
| Alibaba |
qwen-plus |
Use customModel for any model not in the preset list, e.g. anthropic/claude-sonnet-4 via OpenRouter.
Supported API Services
Any service that implements the OpenAI Chat Completions API format:
| Service |
Base URL Example |
| OpenAI |
https://api.openai.com/v1 |
| DeepSeek |
https://api.deepseek.com/v1 |
| Moonshot (Kimi) |
https://api.moonshot.cn/v1 |
| Zhipu (GLM) |
https://open.bigmodel.cn/api/paas/v4 |
| Tongyi Qianwen |
https://dashscope.aliyuncs.com/compatible-mode/v1 |
| Any proxy |
https://your-proxy.com/v1 |
How It Works
- Reads staged diff (falls back to working tree diff if none staged)
- Sends diff to the configured API endpoint with a structured prompt
- Writes the generated message into the SCM input box
The diff is truncated to maxDiffLength characters if it exceeds the limit.
Development
git clone https://github.com/feiren200/commitcraft.git
cd commitcraft
npm install
npm run compile
Press F5 in VSCode to launch the Extension Development Host.
Packaging
npm install -g @vscode/vsce
vsce package
This generates a .vsix file that can be installed via "Extensions → Install from VSIX".
Contributing
See CONTRIBUTING.md.
License
MIT
