English | 日本語

A macOS app that automatically corrects English grammar in your Claude Code prompts.

• Real-time Grammar Correction - Automatically analyzes your prompts as you use Claude Code
• 0-100 Scoring System - Get a score for each prompt with color-coded feedback (green/yellow/red)
• Detailed Explanations - See what's wrong and why, with specific corrections for each error
• Improvement Advice - Receive suggestions on how to write better English
• Interactive Chat - Ask follow-up questions about any correction to learn more
• Multi-Provider Support - Choose between Claude, Gemini, or OpenAI as your AI backend
• 7 Response Languages - Get explanations in English, Japanese, Spanish, French, German, Chinese, or Korean
• Automatic Hook Setup - One-click installation of Claude Code hooks

1. Download the latest .dmg from Releases
2. Open the DMG and drag CCLangTutor.app to Applications
3. Launch the app and click "Install Hooks" when prompted

1. Open CCLangTutor.app
2. Go to Settings (⌘,)
3. Select your preferred AI provider (Claude, Gemini, or OpenAI)
4. Enter your API key
5. Click "Save"

The API key is stored securely in macOS Keychain.

You can choose the language for correction explanations and chat responses:

• English (default)
• 日本語 (Japanese)
• Español (Spanish)
• Français (French)
• Deutsch (German)
• 中文 (Chinese)
• 한국어 (Korean)

Note: The target language for corrections is always English. This setting only affects the language of explanations.

If you prefer to configure hooks manually, add the following to ~/.claude/settings.json:

{
  "hooks": {
    "UserPromptSubmit": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "/Applications/CCLangTutor.app/Contents/MacOS/notifier"
          }
        ]
      }
    ]
  }
}

1. Launch CCLangTutor.app (it runs in the background)
2. Use Claude Code as normal
3. Your prompts will be automatically analyzed and corrections displayed in the app
4. Click on any correction to see details
5. Use the chat to ask follow-up questions about the correction

When you use slash commands (e.g., /commit message here), only the argument portion is corrected, not the command itself.

What gets corrected:

• Regular prompts typed in the main input (when Claude is idle)
• Slash command arguments (e.g., /commit fix the bug → "fix the bug" is corrected)

What does NOT get corrected:

• Messages sent while Claude Code is processing (interrupt messages)
• Responses to AskUserQuestion tool (including "Other" custom text)
• Selecting predefined options (these are not user-written text anyway)

This is a limitation of Claude Code's hook system - the UserPromptSubmit hook only fires for regular prompt submissions.

All your prompts are sent to the LLM provider you select (Claude, Gemini, or OpenAI) for grammar correction. If you have concerns about sending your prompts to external services, do not use this app.

Correction history is stored locally in ~/Library/Application Support/CCLangTutor/.

User prompt → Claude Code Hook (UserPromptSubmit)
                    ↓
            notifier CLI
                    ↓
            pending.json (stored)
                    ↓
            CCLangTutor.app (processes)
                    ↓
            AI API → corrections.json
                    ↓
            SwiftUI displays results

• macOS 14.0+
• Xcode 16.0+
• XcodeGen

git clone https://github.com/Saqoosha/CCLangTutor.git
cd CCLangTutor
./scripts/build.sh Release

# The app is located at:
# build/DerivedData/Build/Products/Release/CCLangTutor.app

./scripts/build.sh Debug      # Debug build
./scripts/build.sh Release    # Release build
./scripts/package_dmg.sh      # Package DMG (includes notarization)
./scripts/release.sh 1.0.0    # Release new version

Sources/
├── CCLangTutor/          # Main app (SwiftUI)
├── CCLangTutorCore/      # Shared models
└── notifier/             # CLI hook

MIT