ChitChat - AI Prompt Enhancer

A Chrome extension that enhances prompt writing on ChatGPT using AI analysis. ChitChat provides real-time feedback, suggestions, and one-click improvements to make your prompts more effective.

Features

• Real-Time Prompt Analysis: Automatically rates your prompt strength (0-100) as you type with a 2-second delay
• Visual Progress Bar: Color-coded strength indicator (red to green) with smooth gradients
• Detailed Analysis: Get specific suggestions and an improved version of your prompt
• One-Click Apply: Instantly replace your prompt with the AI-improved version
• Multiple AI Providers: Support for local Ollama, Google AI, OpenAI, and Claude
• Manual & Auto Modes: Toggle between automatic analysis and manual click-to-analyze
• Seamless UI: Floating glassmorphism design that expands from the bottom-right corner

How it Works

ChitChat injects a floating UI element on chatgpt.com that monitors your prompt input. When you stop typing for 2 seconds, it automatically sends your prompt to the configured AI provider for analysis. The extension displays a strength score and provides detailed suggestions for improvement.

AI Providers Supported

• Ollama (Local): Run AI models locally on your machine (default)
• Google AI (Gemini): Cloud-based AI analysis
• OpenAI (GPT): GPT-4o-mini and other OpenAI models
• Claude (Anthropic): Claude 3 Haiku and other Claude models

Installation

Prerequisites

If using Ollama (recommended):

1. Install Ollama from ollama.ai
2. Download a model: ollama pull phi3:mini (2GB) or ollama pull llama3.2:3b (2GB)
3. Start Ollama: ollama serve

Extension Installation

1. Clone this repository or download the source code
2. Open Chrome and navigate to chrome://extensions/
3. Enable Developer mode (toggle in top-right)
4. Click Load unpacked and select the ChitChat directory
5. Navigate to https://chatgpt.com to see ChitChat in action

Configuration

Edit config.js to configure your preferred AI provider:

const CONFIG = {
    API_PROVIDER: 'ollama', // 'ollama', 'google', 'openai', 'claude'
    
    // Ollama Configuration (Local)
    OLLAMA_API_URL: "http://localhost:11434/api/generate",
    OLLAMA_MODEL: "phi3:mini",
    
    // Add your API keys for cloud providers
    GOOGLE_API_KEY: "your-google-api-key",
    OPENAI_API_KEY: "your-openai-api-key",
    CLAUDE_API_KEY: "your-claude-api-key"
};

Usage

1. Auto-Analysis Mode (default): Type in ChatGPT and see real-time strength feedback
2. Manual Mode: Click the progress bar to trigger analysis on demand
3. Detailed Analysis: Click the ✨ button and select "Get Detailed Analysis"
4. Apply Improvements: Use "Apply This Prompt" to replace your input with the improved version

Project Structure

ChitChat/
├── manifest.json          # Extension configuration (Manifest V3)
├── config.js              # API keys and provider settings
├── prompts/
│   └── prompts.js         # AI prompt templates
├── js/
│   └── content.js         # Main extension logic
├── css/
│   └── style.css          # UI styling (glassmorphism design)
├── popup.html             # Extension popup interface
├── js/
│   └── popup.js           # Popup functionality
└── images/                # Extension icons

Local AI Setup (Ollama)

Recommended Models

• phi3:mini (2GB): Fast, good quality for prompt analysis
• llama3.2:3b (2GB): Balanced performance and quality
• llama3.2:8b (5GB): Higher quality, slower processing

Commands

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Download a model
ollama pull phi3:mini

# Start Ollama service with CORS enabled (REQUIRED for Chrome extension)
OLLAMA_ORIGINS="*" OLLAMA_HOST="0.0.0.0:11434" ollama serve

# Test the API
curl -X POST http://localhost:11434/api/generate \
  -H "Content-Type: application/json" \
  -d '{"model": "phi3:mini", "prompt": "Hello", "stream": false}'

Troubleshooting

• CORS Errors: If you see "Access to fetch at 'http://localhost:11434/api/generate' has been blocked by CORS policy", run:

  # Stop any running Ollama instance
  pkill ollama
  
  # Start Ollama with CORS enabled (REQUIRED)
  OLLAMA_ORIGINS="*" OLLAMA_HOST="0.0.0.0:11434" ollama serve

• No Analysis: Check that your AI provider is configured and running
• Slow Responses: Try a smaller model or adjust the timeout in config.js

Development

To modify prompt templates, edit prompts/prompts.js:

const PROMPTS = {
    "strengthPrompt": "Rate this prompt 0-100: {{PROMPT}}",
    "suggestionsPrompt": "Suggest improvements for: {{PROMPT}}",
    "fixPrompt": "Rewrite this prompt better: {{PROMPT}}"
};

License

This project is open source and available under the MIT License.