Glippy GEO Connector for Claude

Check how ready any public website is for AI search engines and AI agents, straight from a Claude conversation. The connector is free, needs no account, sign-in or API key, and only reads public websites.

Connector URL

https://mcp.glippy.dev/mcp

Overview

Name a website and Glippy returns:

An overall GEO (Generative Engine Optimization) score and grade
16 scored categories, including structured data, semantic HTML, machine readability, entity and authority signals, citability and content freshness (see GEO Scoring Categories)
Which AI crawlers the site's robots.txt blocks or allows
Whether the site publishes a sitemap and llms.txt
A prioritised list of fixes, which you can ask Claude to turn into drafts

You can also compare up to eight sites side by side, for example your site against competitors.

Connect in Claude

From the Connectors Directory

Open Settings > Connectors in Claude, select Browse connectors, search for Glippy and select Connect. No sign-in step follows, because the connector has no accounts.

As a custom connector

In Claude, open Settings > Connectors and select Add custom connector.
Name it Glippy GEO and enter the URL https://mcp.glippy.dev/mcp.
Leave the OAuth fields empty and select Add.

On a Team or Enterprise plan, an organisation owner may need to add the connector first in the organisation's connector settings.

Use it in a conversation

Start a new conversation, check that Glippy GEO is switched on in the tools menu, and ask about a website. Claude picks the right tool by itself, so you don't need to name tools.

Example Prompts

Check the AI readiness of glippy.dev and show the top fixes.

Give me a quick GEO score for https://stripe.com/pricing

Compare vercel.com and netlify.com for AI search readiness.

Which AI crawlers does nytimes.com block in its robots.txt?

Does anthropic.com publish an llms.txt file?

A full URL is fine as input. Glippy reduces it to the domain and analyses that domain's homepage.

Tools

All five tools are read-only. They fetch public files and report on them, and never change anything on a website.

Tool	Input	What it returns
`analyze_domain`	One domain	Full report: overall score and grade, 16 category scores, AI-crawler rules, sitemap and llms.txt discovery, and prioritised recommendations.
`get_geo_summary`	One domain	Quick score: overall score and grade, top strengths and the three most important issues.
`compare_domains`	2 to 8 domains	Sites ranked by score, a per-category comparison showing who leads each category, and the top issues for each site.
`check_robots_txt`	One domain	Which AI crawlers (training crawlers, search crawlers and user-triggered fetchers) robots.txt allows or blocks. This is the stated policy, not a test of what the site serves each crawler.
`check_llms_txt`	One domain	Whether the site publishes `llms.txt`, and its contents if it does. Informational only: major AI models do not read llms.txt yet.

How Glippy Fetches

Every request identifies as GlippyBot: Mozilla/5.0 (compatible; GlippyBot/1.0; +https://www.glippy.dev/bot).
Each analysis starts with robots.txt. If it disallows GlippyBot, Glippy stops and tells you why. If robots.txt answers with a server error or not at all, Glippy also stops.
Otherwise it reads only the homepage and a fixed set of public files: sitemap.xml, llms.txt, llms-full.txt and standard /.well-known/ discovery documents. It follows no links and crawls no other pages.
It never signs in, submits forms or changes anything on a site, and it does not run JavaScript.

Site owners can find more detail, and how to block Glippy, on the GlippyBot info page. The same hosted service also powers the Glippy app in ChatGPT.

Limits

Homepage only. The connector analyses a site's homepage, not every page.
Per-domain rate limit. To protect the sites it reads, a domain gets at most three fresh analyses in any ten minutes, however many people ask. Results are cached for ten minutes, so asking again about the same site is usually instant.
Public sites only. Pages behind a login, a paywall or bot protection can't be analysed.
Static HTML. The connector reads the HTML the server sends. Content that only appears after JavaScript runs is not seen, which is also how many AI crawlers see a site.

Privacy

The connector runs on Glippy's servers on Google Cloud Run in the EU (Netherlands). The only data it needs is the domain name or names you ask about. It does not receive your name, email address or Claude conversation, and it doesn't use any data for tracking, advertising or training AI models. Read the full privacy policy.

Troubleshooting

Message	What it means
The site's robots.txt disallows GlippyBot	The site owner has opted out of Glippy. This is intended. If it is your own site, allow `GlippyBot` in robots.txt (see the GlippyBot info page).
Analysed several times in the last few minutes	The per-domain rate limit was reached. Try again after the number of minutes shown, or check a different domain.
The site refused GlippyBot's request (HTTP 401, 403, 429 or 503)	A firewall or bot-protection service blocked the request. Site owners can allow-list GlippyBot in their WAF.
Glippy is handling a lot of requests right now	The service is busy. Try again in a minute.

Connector vs Glippy MCP Server

The connector is a free, quick way to check a site's homepage. The Glippy MCP server is a paid tool that runs on your own machine, for deeper audits:

	Claude connector	Glippy MCP server
Price	Free, no account	License key
Runs on	Glippy's servers	Your machine (Node.js)
Scope	Homepage per domain	Multi-page crawls, sitemaps and URL lists
Extras	–	Agent access check, JavaScript rendering with Chrome, HTML and bulk report exports

Setup for the MCP server is in Getting Started.

Support

Questions, bugs or a site that doesn't analyse as expected: email [email protected] or see the support page.