PromptBooster

A VS Code extension that enhances and optimizes your prompts using AI before sending to GitHub Copilot.

🎯 Features

PromptBooster automatically detects your context to provide the best optimization experience:

• Chat: Type @PromptBooster to instantly optimize your chat prompt.
• Editor: Right-click any .prompt.md file to boost it.

✅ Real-time Optimization (Chat)

• Context-Aware: Reads the content of files you reference (e.g., #file:utils.ts) to give the AI crucial context.
• Smart Intent Detection: Automatically detects if you are Asking or Editing.
• Dynamic Actions:
  • $(sparkle) Apply to Chat: Copies the optimized prompt to Copilot input.
  • $(new-file) Refine in File: Creates a .prompt.md file for manual editing.

✅ Manual Refinement (Editor)

• Persistent: Work in .prompt.md files for complex prompt engineering.
• Version Control: Save your best prompts to .github/prompts/.
• Right-click Boost: Enhance your drafts with one click.
• Process Prompt File: Execute your perfected prompt directly to Copilot via the Status Bar or Command Palette.

🚀 Quick Start

Installation

1. Install "PromptBooster" from VS Code Extensions
2. Ensure GitHub Copilot extension is installed and active

1. In Chat (Real-time)

1. Open Copilot Chat.
2. Type @PromptBooster #file:current.ts Refactor this code.
3. Intent Detection: The extension sees you want to Edit.
4. Click $(sparkle) Apply to Chat to use the optimized prompt.

2. In Editor (Manual File)

1. Create a file named my-task.prompt.md.
2. Write your rough idea:

   Write a react component for a login form
   

3. Right-click → PromptBooster: Boost This Prompt.
4. The file updates with a professional, structured prompt.
5. Click the Run CodeLens or "Process Prompt" in the status bar to send it to Copilot.

📦 Building & Packaging

To build the extension for distribution:

npm run package

This generates a .vsix file in the dist/ folder, which is automatically excluded from git. You can then:

• Test locally: Install the VSIX manually in VS Code
• Publish: Use vsce publish to publish to the VS Code Marketplace
• Share: Distribute the VSIX file to others for manual installation

⚙️ Configuration

Configure in VS Code Settings (Cmd+,):

📋 Commands

Access via Command Palette (Cmd+Shift+P) or use npm scripts:

VS Code Commands

Access via Command Palette (Cmd+Shift+P):

• PromptBooster: Boost This Prompt - Enhance .prompt.md file (Manual Mode)
• PromptBooster: Switch Operation Mode - Change between modes
• PromptBooster: Toggle Auto-Optimization - Turn auto-optimization on/off
• PromptBooster: Switch AI Model - Choose which AI model to use
• PromptBooster: Configure Permissions - Manage interception permissions
• PromptBooster: Process Prompt File - Process generated .prompt.md file (File Mode)
• PromptBooster: Test File Generation - Test file generation feature
• PromptBooster: Test Realtime Mode - Test realtime mode integration

🛠️ Requirements

Runtime

• VS Code: 1.99.0 or higher
• GitHub Copilot: Active subscription (for Language Model API access)
• OS: macOS, Windows, or Linux

Development & Building

• Node.js: 20.x or higher
• npm: 8.x or higher
• VSCE: Automatically installed with dev dependencies (for npm run package)

🎯 What's Implemented

🔧 Manual Mode

• Right-click enhancement of .prompt.md files
• AI-powered prompt optimization and structuring
• Flexible AI model selection (gpt-4.1, gpt-4o, claude-haiku-4.5)
• Selection-based partial optimization
• Real-time progress notifications with cancel option

⚡ Real-time Mode

• Chat participant integration (@PromptBooster in Copilot chat)
• Automatic prompt interception and enhancement
• Smart Intent Detection - Distinguishes between "ask" and "edit" intents
• Dynamic context-aware button suggestions
• Structured prompt format with Task/Context/Requirements/Output sections
• Chat reference support (#selection, #file, #editor)
• Direct AI response streaming
• Auto-optimization with configurable timeout

📝 File Mode

• Automatic .prompt.md file generation from chat input
• Three file naming strategies: timestamp, prompt-based, custom
• Dedicated file storage in .github/prompts/ directory
• Collision detection and automatic resolution
• Process button for direct execution
• CodeLens indicators for quick workflow
• HTML comment metadata preservation
• Full editor integration with syntax highlighting

🛠️ Core Infrastructure

• Comprehensive configuration system with VS Code settings integration
• Mode management and switching via status bar
• Language model provider abstraction with fallback logic
• Dependency injection container for service orchestration
• Robust error handling and logging
• Full extension API compliance

🚀 How It Works

Manual Mode

flowchart TD
    A[Right-click .prompt.md] --> B{Valid File?}
    B -->|Yes| C[Select AI Model]
    C --> D[Optimize Prompt]
    D --> E[Update File Content]

Real-time Mode (Smart Intent)

flowchart TD
    Start[User Input @PromptBooster] --> Intercept[Intercept & Optimize]
    Intercept --> Check{Detect Intent}
    
    Check -->|Edit Code| EditUI[Show 'Apply Edits' / 'Refine in File']
    Check -->|Ask Question| AskUI[Show 'Ask Copilot' / 'Edit']
    
    EditUI -->|User Click| Run[Execute Optimized Prompt]
    AskUI -->|User Click| Run

File Mode

flowchart TD
    Start[User Chat Input] --> Gen[Generate .prompt.md]
    Gen --> Open[Open in Editor]
    Open --> Edit[Manual Editing]
    Edit --> Process[Click Process Button]
    Process --> Send[Send to Copilot]

🎨 Status Bar & UI

• Status bar item (bottom right): Shows current mode
  • 🔧 Manual Mode - File-based enhancement
  • ⚡ Real-time Mode - Chat interception with preview
  • 📝 File Mode - Generate editable files
• Click to switch modes - Opens Quick Pick menu
• Process button - Appears in file mode for .prompt.md files
• CodeLens - "▶️ Process this prompt" at top of .prompt.md files
• Progress notifications - With cancel button during optimization
• Output channel - Detailed logging in "PromptBooster" channel

💡 Tips

For Best Results

1. Use with complex prompts: Let PromptBooster add structure and detail
2. Review in real-time mode: Learn how prompts are enhanced
3. Use references: #selection, #file provide better context
4. Check the logs: View → Output → PromptBooster shows what's happening
5. Try different models: Use "Switch AI Model" to experiment
6. Edit before processing: File mode gives full control

When to Use Each Mode

File Mode Naming Strategies

🔗 Links

• GitHub Repository
• GitHub Copilot Extension
• VS Code Extension API
• Language Model API Docs

📄 License

MIT License - Created by Son Quach

🐛 Troubleshooting

"No language model available"

• Ensure GitHub Copilot is installed and active
• Sign in to Copilot with your GitHub account
• Check your Copilot subscription status
• Restart VS Code

Context menu doesn't appear

• File must end with .prompt.md (case-sensitive)
• Save the file before right-clicking
• Reload window: Developer: Reload Window

Real-time mode not working

• Enable auto-optimize: Cmd+Shift+P → "Toggle Auto-Optimization"
• Check current mode in status bar (should show ⚡ Real-time)
• Use @PromptBooster prefix in chat
• Check Output channel for errors

File mode not creating files

• Ensure workspace folder is open
• Check fileOutputDirectory setting (default: .github/prompts)
• Verify write permissions in workspace
• Check Output channel for detailed errors

Optimization timeout

• Default timeout is 10 seconds
• Complex prompts may need more time
• Extension falls back to original prompt with context
• Check Output channel for "Optimization timed out" message

Chat references not included

• Use #selection, #file, or #editor in your prompt
• References are automatically extracted and added to context
• Check Output channel for "Including X reference(s)" message
• Verify files are saved before using #file reference

🙋 Support

• Open an issue on GitHub
• Review CHANGELOG.md for recent changes

PromptBooster: Enhance your prompts. Better prompts → Better results! ✨