Auto Commit Gen
AI-optional commit message generation inside VS Code.
Auto Commit Gen reads your staged Git changes, analyzes the diff, and suggests useful commit messages. It works offline with a rule-based generator by default, and can optionally use Ollama, OpenAI, Claude, or hybrid mode for AI-assisted suggestions.
Features
- Rule-based generation by default, with no API key required
- Optional AI generation with Ollama, OpenAI, or Claude
- Hybrid mode that tries AI first and falls back to local rules
- Conventional, simple, detailed, and emoji commit styles
- Preview picker before applying a generated message
- Generate & Commit command with confirmation
- Status bar and window progress while generating
- Helpful error messages with suggested fixes
- Advanced settings for timeout, temperature, message length, and fallback behavior
- Telemetry setting that defaults off; the extension currently sends no telemetry
Commands
| Command |
Description |
Auto Commit Gen: Generate Commit Message |
Generate a message and place it in the Source Control input |
Auto Commit Gen: Generate & Commit |
Generate, preview, confirm, and run git commit |
Auto Commit Gen: Open Settings |
Open this extension's settings |
Keyboard Shortcuts
| Command |
Windows/Linux |
macOS |
| Generate Commit Message |
Ctrl+Alt+G |
Cmd+Alt+G |
| Generate & Commit |
Ctrl+Alt+C |
Cmd+Alt+C |
Quick Start
- Open a Git repository in VS Code.
- Make changes to files.
- Stage your changes in Source Control or with
git add.
- Run
Auto Commit Gen: Generate Commit Message.
- Pick or edit a suggestion.
- Commit normally, or use
Auto Commit Gen: Generate & Commit.
If no staged changes are found, the extension can offer to stage all changes and generate again.
Providers
Rule-Based
Default mode. It does not call any AI service.
"autoCommitGen.provider": "rule-based"
This mode uses file names, file categories, additions, deletions, and change size to produce a deterministic message.
Ollama
Uses a local Ollama server.
"autoCommitGen.provider": "ollama",
"autoCommitGen.ollamaEndpoint": "http://localhost:11434",
"autoCommitGen.ollamaModel": "llama3.2"
Start Ollama before generating:
ollama serve
OpenAI
Uses the OpenAI API.
"autoCommitGen.provider": "openai",
"autoCommitGen.openaiModel": "gpt-4o-mini",
"autoCommitGen.apiKey": "sk-..."
Claude
Uses the Anthropic Claude API.
"autoCommitGen.provider": "claude",
"autoCommitGen.claudeModel": "claude-haiku-4-5-20251001",
"autoCommitGen.apiKey": "sk-ant-..."
Hybrid
Generates a rule-based message first, then tries the selected AI provider. If AI fails or times out, the rule-based message is still available.
"autoCommitGen.provider": "hybrid",
"autoCommitGen.hybridProvider": "ollama"
hybridProvider can be ollama, openai, or claude.
Commit Styles
| Style |
Example |
conventional |
feat(auth): add refresh token support |
simple |
Add refresh token support |
detailed |
feat(auth): add refresh token support plus a body |
emoji |
✨ feat(auth): add refresh token support |
Settings
| Setting |
Default |
Description |
autoCommitGen.provider |
rule-based |
Message generation provider |
autoCommitGen.hybridProvider |
ollama |
AI provider used by hybrid mode |
autoCommitGen.commitStyle |
conventional |
Generated message format |
autoCommitGen.maxDiffLines |
300 |
Maximum diff lines sent to AI |
autoCommitGen.autoStage |
false |
Run git add -A before generating |
autoCommitGen.showPreview |
true |
Show suggestion picker before applying |
autoCommitGen.numSuggestions |
3 |
Number of AI suggestions to request |
autoCommitGen.subjectMaxLength |
72 |
Preferred max length for the first line |
autoCommitGen.aiTemperature |
0.5 |
AI creativity level |
autoCommitGen.aiTimeoutSeconds |
30 |
Max wait time for AI providers |
autoCommitGen.hybridTimeoutSeconds |
20 |
Max wait time before hybrid fallback |
autoCommitGen.confirmBeforeCommit |
true |
Confirm before running git commit |
autoCommitGen.showFallbackWarnings |
true |
Warn when hybrid mode falls back |
autoCommitGen.telemetryEnabled |
false |
Reserved opt-in flag; no telemetry is currently sent |
Open settings from the Command Palette:
Auto Commit Gen: Open Settings
Privacy
Rule-based mode keeps everything local.
Ollama sends the diff only to your local Ollama server.
OpenAI and Claude modes send staged diff content to the selected external API provider. Review your staged changes before using cloud providers if your repository contains private or sensitive code.
Auto Commit Gen does not currently send telemetry. The autoCommitGen.telemetryEnabled setting defaults to false and is included so privacy-conscious users can verify telemetry behavior from VS Code settings.
AI provider code is lazy-loaded. The OpenAI, Claude, and Ollama generation modules are imported only when you run a command with an AI-backed provider.
Local Development
Install dependencies:
npm install
Compile:
npm run compile
Lint:
npm run lint
Run locally:
- Open this folder in VS Code.
- Press
F5.
- In the Extension Development Host window, open a Git repository.
- Stage changes and run one of the Auto Commit Gen commands.
Troubleshooting
No Git repository found
Open a folder that contains a .git directory, or initialize one:
git init
No staged changes found
Stage files first:
git add <file>
Or enable:
"autoCommitGen.autoStage": true
Cannot connect to Ollama
Start Ollama and confirm the endpoint setting:
ollama serve
"autoCommitGen.ollamaEndpoint": "http://localhost:11434"
API key not set or invalid
Set autoCommitGen.apiKey, or switch back to rule-based mode.
AI times out
Try one of these:
- Increase
autoCommitGen.aiTimeoutSeconds
- Lower
autoCommitGen.maxDiffLines
- Use
rule-based provider
- Use
hybrid provider for fallback behavior
Current Status
The extension is functional for local testing. Before publishing to the VS Code Marketplace, replace the placeholder publisher id, add marketplace assets, add a license file, add a changelog, and run real Extension Development Host testing across several repositories.
License
MIT
