AMA (Ask Me Anything) AI - Chrome Extension

A Chrome extension that lets you ask questions about any website you’re visiting, powered by Firecrawl and GPT-4o-mini. This extension crawls web pages, processes their content, and enables natural language interactions about the site’s content.

Install from Chrome Web Store

• 🤖 Ask questions about any website’s content in natural language
• 🌐 Smart web crawling that respects site structure and robots.txt
• 💬 Chat-like interface with streaming responses
• 🧠 Context-aware responses with references to specific pages
• 🔄 Maintains conversation history per domain
• ⚡ Real-time streaming responses from GPT-4o-mini
• ⚙️ Highly configurable crawling parameters
• 🔒 Secure API key management
• 📝 Markdown formatting support in responses
• 🔗 Automatic link references in answers

1. Initial Scan: When you click “Start Crawl”, the extension:

   • Validates your API keys
   • Checks the current domain
   • Initializes a new conversation history
   • Starts the Firecrawl process

2. Progress Tracking:

   • Shows a progress bar with real-time updates
   • Displays number of pages crawled
   • Indicates crawling status and completion

3. Content Processing:

   • Converts HTML to clean markdown format
   • Extracts page titles and URLs
   • Maintains source references for citations
   • Truncates content to stay within API limits

1. Content Preparation:

   • Organizes crawled content by relevance
   • Maintains page structure and hierarchy
   • Preserves source URLs for citations

2. AI Processing:

   • Streams requests to GPT-4o-mini
   • Maintains conversation context
   • Generates responses with source citations

3. Response Handling:

   • Real-time streaming of answers
   • Markdown formatting for readability
   • Automatic link insertion
   • Source reference preservation

• API Key: Your Firecrawl authentication key
• Max Depth (default: 3):
  • Controls how many links deep the crawler will go
  • Higher values explore more nested pages
  • Recommended range: 1-5 for optimal performance

• Page Limit (default: 50):

  • Maximum number of pages to crawl
  • Higher limits allow more comprehensive coverage
  • Consider API usage when adjusting
  • Range: 1-1000 pages

• Allow Backward Links (default: true):

  • When enabled, crawler follows links to previously visited domains
  • Useful for sites with cross-referenced content
  • Disable to stay within single domain

• Timeout (default: 20000ms):

  • Maximum time to wait for each page
  • Prevents hanging on slow-loading pages
  • Recommended range: 5000-30000ms

• Wait For (default: 2000ms):

  • Delay between page requests
  • Helps respect server rate limits
  • Adjust based on site’s robustness
  • Range: 0-5000ms

• Max Content Length (default: 250,000 characters):
  • Maximum characters sent to GPT-4o-mini
  • Balances comprehensiveness with API limits
  • Range: 1-500,000 characters
  • Higher values may increase API costs

• Model Selection:
  • gpt-4o-mini: Faster, more concise responses
  • gpt-4o: More detailed, nuanced answers, much more expensive
  • Choose based on your needs for speed vs. detail and cost

The extension maintains separate conversation histories for each domain:

{
  "example.com": [
    {"role": "user", "content": "What is this site about?"},
    {"role": "assistant", "content": "Based on [Home Page](https://example.com), this site..."},
    // Additional messages...
  ]
}

The extension processes content in multiple stages:

1. HTML Processing:

   // Example of content processing
   {
     "title": "Page Title",
     "url": "https://example.com/page",
     "content": "Processed markdown content...",
     "metadata": {
       "sourceURL": "https://example.com/page",
       "crawlTime": "2024-01-01T00:00:00Z"
     }
   }
   

2. Content Organization:

   • Groups related content
   • Maintains hierarchy
   • Preserves source references

Responses are streamed in real-time:

// Example streaming response format
{
  "role": "assistant",
  "content": "According to [About Page](https://example.com/about)...",
  "references": [
    {"title": "About Page", "url": "https://example.com/about"},
    // Additional references...
  ]
}