An AI-powered prompt optimization tool built on Val Town that iteratively improves prompts using multiple specialized AI workers.

• Multi-stage Optimization: Uses three specialized workers (Clarifier, Condenser, Includer) to improve prompts
• Iterative Refinement: Processes prompts through multiple cycles until optimal results are achieved
• Final Organization: Applies a final organization pass to structure the improved prompt
• Modern UI: Clean, responsive interface built with TailwindCSS
• Val Town Integration: Fully compatible with Val Town's serverless platform

The application consists of:

1. Main HTTP Handler (backend/index.http.ts): Serves the frontend and API endpoint
2. Worker Functions: Three specialized prompt improvement functions
   • Clarifier: Makes prompts clearer and more specific
   • Condenser: Removes redundancy while maintaining meaning
   • Includer: Adds missing context and relevant details
3. Frontend: Single-page application with real-time prompt improvement

• Val Town account (paid subscription recommended for API features)
• OpenAI API key configured in Val Town environment variables

1. Go to Val Town
2. Create a new HTTP val named prompt_improver
3. Copy the contents of backend/index.http.ts to your val
4. Save and deploy

In your Val Town settings, ensure you have:

• OPENAI_API_KEY - Your OpenAI API key
• valtown or VAL_TOWN_API_KEY - Automatically set by Val Town

Your application will be available at:

https://[your-username]-prompt-improver.web.val.run

POST https://[your-username]-prompt-improver.web.val.run/api/improve

{
  "prompt": "Your prompt to improve"
}

{
  "result": "The improved prompt",
  "steps": 5,
  "history": [
    {
      "step": 1,
      "worker": "Clarifier",
      "prompt": "Clarified version..."
    }
  ]
}

1. Install Deno
2. Run the application:

   deno run --allow-net --allow-env backend/index.http.ts

3. Open http://localhost:8000

Create a .env file:

OPENAI_API_KEY=your_openai_api_key
VAL_TOWN_API_KEY=your_valtown_api_key

1. Input: User provides a prompt to improve
2. Processing: The prompt goes through multiple optimization cycles:
   • Each cycle applies three workers in sequence (Clarifier → Condenser → Includer)
   • Process continues until the prompt stabilizes or reaches maximum iterations
3. Organization: A final pass organizes the improved prompt for clarity
4. Output: Returns the optimized prompt with processing statistics

The MAX_STEPS constant in the backend controls the maximum number of optimization cycles (default: 10).

The application uses OpenAI's gpt-4o-mini model by default. You can change this in the worker functions.

Each worker has different temperature settings optimized for its task:

• Clarifier: 0.3 (more focused)
• Condenser: 0.2 (very focused)
• Includer: 0.4 (slightly creative)
• Organizer: 0 (deterministic)

1. "Failed to improve prompt" error

   • Check your OpenAI API key is correctly configured
   • Ensure you have API credits available
   • Check Val Town logs for detailed error messages

2. Val not found errors

   • Ensure the val is deployed with the correct name
   • Check that your Val Town account has the necessary permissions

3. CORS errors

   • The application includes CORS headers by default
   • If issues persist, check your browser's security settings

Enable detailed logging by checking the Val Town logs:

1. Go to your val's page on Val Town
2. Click on the "Logs" tab
3. Review any error messages or console outputs

Feel free to fork this project and submit improvements. Key areas for enhancement:

• Additional worker types
• Support for different AI models
• Batch processing capabilities
• Prompt history and favorites

MIT License - Feel free to use and modify as needed.

For issues or questions:

• Create an issue in this repository
• Contact through Val Town community forums
• Check Val Town documentation at https://docs.val.town