NEW Feature: Visit your customized dashboard and sync all your results on the cloud.
Analyze 1 page
Get an instant LLM visibility score and know exactly what to fix.
Compare 2 pages
Analyze two pages simultaneously and see exactly where you win, lose, or tie across every pillar.
Website Audit
Crawl your entire sitemap and get an aggregate LLM visibility score for every page.
Scan History
Track your scores over time and measure the impact of your optimizations.
What questions people ask
Discover real questions people are asking about any topic, powered by AI.
AI Crawler Tester
Test AI crawlers against your pages and see who is blocked and who gets through.
llms.txt Generator
Generate a compliant llms.txt file that helps AI systems understand your site at a glance.
robots.txt Generator
Control which AI crawlers can access your content with a properly configured robots.txt.
JSON-LD Schema Generator
Build structured data markup that gives LLMs explicit signals about your content's meaning.
Meta Tags Generator
Generate complete, optimized meta tags including Open Graph and canonical for any page.
MCP Connector
Connect Claude and other AI agents to the hey-eye analyzer and run visibility checks straight from the chat.
Blog
AI visibility insights, optimization strategies, and practical guides for LLM-ready content.
Support
Guides, tutorials, and documentation for hey-eye.
Structural Integrity 30%
Title, headings, canonical, Open Graph, semantic HTML and language attributes.
AI Extractability 35%
JSON-LD schema, paragraph length, lists, date signals, internal links and breadcrumbs.
Content Clarity 15%
Sentence length, heading density and Flesch readability — with Greek language support.
Authority & Trust 20%
Author attribution, social profiles, image alt text, meta robots and robots.txt AI access.
Analyze 1 page Compare 2 pages Website Audit Scan History
Questions Asked AI Crawler Tester llms.txt Generator robots.txt Generator JSON-LD Schema Generator Meta Tags Generator
MCP Connector Blog Support Structural Integrity AI Extractability Content Clarity Authority & Trust
Login Dashboard Contact

How to Use the hey-eye MCP Connector

Connect hey-eye to Claude or ChatGPT and run AI visibility analysis, page comparisons, and fixes directly in your chat.

What is the MCP Connector?

The hey-eye MCP Connector lets you run AI visibility analysis inside your favorite LLM.

No tab switching. No copy-pasting URLs into a separate tool. You type a prompt, the LLM calls hey-eye, and results appear right in your conversation.

It works with Claude and ChatGPT.

The connector URL:

https://hey-eye.gr/mcp

What You Can Do

Analyze a Page

Ask your LLM to analyze any public URL. hey-eye returns an overall AI visibility score plus a breakdown across four pillars:

  • Structural Integrity (30%) — heading hierarchy, HTML semantics, internal linking
  • AI Extractability (35%) — schema markup, content structure, summary patterns
  • Content Clarity (15%) — readability, definitional patterns, reduced ambiguity
  • Authority & Trust (20%) — author attribution, trust signals, source credibility

Example prompt:

Analyze the AI visibility of https://example.com/pricing

You get a score for each pillar, the overall score, and specific issues to fix.

Compare Two Pages

Want to benchmark against a competitor? Ask the LLM to compare two URLs. hey-eye scores both pages across all four pillars and shows you exactly where you win, lose, or tie.

Example prompt:

Compare the AI visibility of https://mysite.com/feature and https://competitor.com/feature

The side-by-side breakdown tells you which structural or content gaps to close first.

Test AI Crawler Access

Scoring a page is only half the story. If AI crawlers cannot reach it in the first place, the score does not matter.

Ask your LLM to test whether AI bots can actually fetch a URL. hey-eye checks 11 major AI crawlers on two separate layers:

  • robots.txt: what the site’s rules say each bot is allowed to do
  • Live request: what actually happens when the page is fetched with that bot’s real user-agent, catching firewall and bot-protection blocks that robots.txt alone would hide

Bots covered: GPTBot, ChatGPT-User, ClaudeBot, Claude-Web, PerplexityBot, Google-Extended, Applebot-Extended, CCBot, Bytespider, cohere-ai, Amazonbot.

Example prompt:

Test AI crawler access for https://example.com

You get a per-bot verdict:

  • allowed: the bot can reach the page
  • blocked_robots: blocked by robots.txt only
  • blocked_firewall: allowed by robots.txt, but blocked by the firewall anyway
  • blocked_both: blocked on both layers

The blocked_firewall verdict is the one to watch. It flags bots you think you are allowing, but your network is silently blocking. That gap is invisible in robots.txt, and it is a common reason a page never shows up in AI search.

Example follow-up:

ClaudeBot shows blocked_firewall. What are the likely causes and how do I allow it?

Fix Issues in the Chat

This is where it gets powerful.

After an analysis, you can ask the LLM to fix the issues right there in the conversation. The LLM already has the full score breakdown, so it knows exactly what failed and why.

Example follow-ups:

The JSON-LD schema is missing. Generate the correct Article schema for this page.

My heading hierarchy has gaps. Suggest a corrected H1-H2-H3 structure.

Write a summary box I can add above the fold to improve extractability.

No context switching. The analysis and the fix happen in the same thread.

Setup: Claude

  1. Open Claude (claude.ai)
  2. Go to Settings (bottom left)
  3. Click Connectors
  4. Click Add Connector and paste the URL: 
    https://hey-eye.gr/mcp
  5. Approve access when prompted
  6. Start a new chat and ask Claude to analyze a URL

That’s it. Every chat can now use hey-eye tools.

Setup: ChatGPT

  1. Open ChatGPT (chatgpt.com)
  2. Start a new chat
  3. Click the connectors/tools icon in the input bar
  4. Select Add MCP server and paste the URL: 
    https://hey-eye.gr/mcp
  5. Approve access when prompted
  6. Ask ChatGPT to analyze or compare a URL

ChatGPT will call the hey-eye tools automatically when your prompt mentions AI visibility, scoring, or page analysis.

Tips for Better Results

Be specific with your prompts. “Analyze https://example.com” works. But “Analyze the AI visibility of https://example.com and list the top 3 issues to fix” gets you more actionable output.

Use compare for competitive intel. Pick a competitor page that targets the same keyword. The pillar-by-pillar comparison reveals gaps you can close in an afternoon.

Chain analysis with action. Run the analysis first. Then ask the LLM to generate the missing schema, rewrite a heading structure, or draft a summary box. The LLM uses the score context to make targeted fixes.

Re-analyze after changes. Once you push fixes live, run the analysis again from the same chat. You can track your score improvement without leaving the conversation.

Check access before you optimize. Run a crawler access test first. There is no point improving a page’s score if AI bots are blocked from reading it. Fix access, then fix content.

What the Connector Does Not Do

  • It does not crawl entire sitemaps. For bulk audits across many pages, use the Website Audit on hey-eye.gr directly.
  • It does not store your results. Each analysis is ephemeral. To save scores over time, use the hey-eye Dashboard.
  • It does not require an API key or paid plan. The MCP connector is available to all users.

Troubleshooting

“Tool not found” or no response from hey-eye Make sure you pasted the full URL including https://. Disconnect and reconnect the connector if the issue persists.

Scores look different from the website The connector runs the same scoring engine as hey-eye.gr. If you see a difference, check that you are analyzing the exact same URL (with or without trailing slash, www vs non-www).

Rate limiting The connector has per-request rate limiting. If you get a 429 error, wait a few seconds and try again. Avoid rapid-fire analysis of dozens of URLs in a single chat.

A bot shows “blocked_firewall” but I never blocked it This is not a connector error. It means the site’s robots.txt allows that bot, but a network layer (firewall, CDN, or bot-protection service like Cloudflare Bot Management) is blocking the live request anyway. The block is real and lives on the site’s infrastructure, not in hey-eye. Check your WAF rules, bot-fighting settings, or security level to allow the crawlers you want in.

Crawler test shows blocks, but the page loads fine in my browser Bots and browsers are treated differently. A page can be fully open to human visitors and still block AI crawlers by user-agent. The crawler test fetches each page with the bot’s real user-agent, so it sees exactly what the crawler sees, not what you see.