🚀 Hyper-Reader: The Agentic Web Bridge

Stop feeding your LLM messy HTML. Hyper-Reader delivers high-fidelity, ad-free content optimized for Claude, GPT-4, and Gemini with sub-second response times.

Built by Jason Pellerin AI Solutionist — the same engineering behind enterprise AI voice agents and automation systems.

---

Why Hyper-Reader?

Problem	Hyper-Reader Solution
Raw HTML is noisy and token-expensive	Clean Markdown with smart content extraction
Anti-bot systems block your scrapers	3-level stealth with fingerprint randomization
Different LLMs need different formats	Agent-optimized presets (Claude, GPT, Gemini)
Cold starts kill your agent's speed	Standby Mode for 1-second responses
Single pages lack context	Deep Read follows links for comprehensive data

---

🎯 Agent Presets

Choose your target LLM for optimized output:

Claude (Default)

<document>
  <metadata>
    <title>Article Title</title>
    <author>John Doe</author>
    <published>2024-01-15</published>
    <source>https://example.com/article</source>
  </metadata>
  <content>
    # Main Heading
    
    Clean, structured Markdown content...
  </content>
</document>

GPT-4

# Article Title

> Source: https://example.com/article
> Author: John Doe | Published: 2024-01-15

Content with inline citations [1] and reference links...

---
## References
[1]: https://example.com/article "Original Source"

Gemini

Compact Markdown optimized for Gemini's context window with aggressive token optimization.

SearchGPT

Web-search optimized format with prominent source attribution and fact-checkable structure.

---

⚡ Standby Mode

Enable Standby Mode for instant API responses. Your Actor stays warm and ready:

# Response time: ~1 second vs 30+ seconds cold start
curl -X POST "https://YOUR_ACTOR_STANDBY_URL/extract" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com", "agentPreset": "Claude"}'

Perfect for:

• Real-time AI assistants
• MCP tool integrations
• Cursor/Claude Desktop extensions
• n8n and automation workflows

---

🛡️ Stealth Levels

Level 1: Basic

• Standard datacenter proxies
• Basic header rotation
• Best for: Blogs, news sites, documentation

Level 2: Standard (Default)

• Residential proxy rotation
• Browser fingerprint randomization
• WebGL/Canvas spoofing
• Best for: E-commerce, social media, most protected sites

Level 3: Elite

• Premium residential proxies
• Human-like mouse movements
• Session persistence
• Full anti-fingerprinting
• Best for: LinkedIn, Amazon, heavily protected sites

---

🔍 Deep Read

Gather comprehensive context by following internal links:

{
  "url": "https://example.com/product",
  "deepReadDepth": 2,
  "deepReadMaxPages": 10
}

Returns aggregated content from the main page plus related pages (About, FAQ, Reviews, etc.) in a single, structured document.

---

📸 Vision Screenshots

Capture page screenshots for Vision model analysis:

{
  "url": "https://example.com",
  "useVision": true
}

Returns a 1280x720 optimized PNG stored in Apify's Key-Value Store, perfect for GPT-4V, Claude Vision, or Gemini Pro Vision.

---

Input Schema

Field	Type	Default	Description
url	string	-	Target URL to extract
urls	array	-	Multiple URLs (batch mode)
agentPreset	enum	Claude	Output optimization target
outputFormat	enum	markdown	markdown, json, or html_cleaned
stealthLevel	integer	2	1-3 (Basic to Elite)
useVision	boolean	false	Capture screenshot
deepReadDepth	integer	0	Link following depth (0-3)
waitForSelector	string	-	CSS selector to wait for
excludeSelectors	string	-	Elements to remove (comma-separated)
maxContentLength	integer	0	Truncate output (0 = unlimited)

---

Output Structure

{
  "url": "https://example.com/article",
  "finalUrl": "https://example.com/article/",
  "format": "markdown",
  "agentPreset": "Claude",
  "content": "# Article Title\n\nClean markdown content...",
  "metadata": {
    "title": "Article Title",
    "author": "John Doe",
    "publishDate": "2024-01-15",
    "description": "Article description...",
    "wordCount": 1500,
    "readingTimeMinutes": 7
  },
  "screenshotUrl": "https://api.apify.com/v2/key-value-stores/.../screenshot.png",
  "processingTimeMs": 2340,
  "charCount": 8500,
  "extractedAt": "2024-01-15T10:30:00.000Z"
}

---

Use Cases

🤖 AI Agent Research

// Feed clean web data to your AI agent
const result = await client.call('ai_solutionist/hyper-reader', {
  url: 'https://docs.example.com/api',
  agentPreset: 'Claude'
});
// result.content is ready for your LLM context

📊 Competitive Intelligence

// Extract competitor pages with deep context
const result = await client.call('ai_solutionist/hyper-reader', {
  url: 'https://competitor.com/pricing',
  deepReadDepth: 2,
  agentPreset: 'GPT-4'
});

🔗 MCP Tool Integration

{
  "mcpServers": {
    "hyper-reader": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/mcp-apify"],
      "env": {
        "APIFY_TOKEN": "your_token",
        "ACTOR_ID": "ai_solutionist/hyper-reader"
      }
    }
  }
}

📰 News Aggregation

// Batch extract multiple articles
const result = await client.call('ai_solutionist/hyper-reader', {
  urls: [
    'https://news.site/article1',
    'https://news.site/article2',
    'https://news.site/article3'
  ],
  agentPreset: 'Gemini',
  outputFormat: 'json'
});

---

Pricing

Tier	Price	Features
Standard	$1 / 1,000 pages	Full extraction, all presets, Stealth 1-2
Elite	$5 / 1,000 pages	Stealth Level 3, residential proxies
Pro Monthly	$49 / month	Standby Mode, unlimited standard proxy

---

Support

• Documentation: jasonpellerin.com/hyper-reader
• Issues: Open an issue on GitHub
• Enterprise: Contact Jason Pellerin

---

Built with 🔥 by Jason Pellerin AI Solutionist

Transforming web chaos into agent-ready intelligence.

Build timestamp: Sun Jan 18 16:29:53 MST 2026