How Webloyt works

Webloyt is free hosting for websites your AI builds. You describe the site, your AI writes the HTML and CSS, and Webloyt checks it and publishes it at your-name.webloyt.cloud: the harness around the AI, so to speak. This page covers what you can publish, what you can't, and what Webloyt does for you automatically.

Getting started

1. In Claude, add Webloyt from the connector directory. In another AI app (for example Mistral), add it as a custom connector with this address:
   https://webloyt-mcp.iroiro.cloud/mcp
2. Sign in with Google when asked. Webloyt only receives your email address.
3. Ask for a website. For example:

Website names are 7–63 characters: lowercase letters, digits and hyphens. Your AI handles everything else: choosing file names, uploading, fixing problems Webloyt reports back, and giving you the live link.

What your AI can do

Limits

Allowed and not allowed

Webloyt sites are static: HTML and CSS only. That keeps them fast, safe for visitors, and free to host.

If something isn't allowed, Webloyt rejects the upload with a message that names the exact problem, so your AI can fix it and try again. Nothing is published until it passes.

Images, fonts and stylesheets

Images

Use images uploaded to your site, or direct HTTPS image addresses from anywhere, such as Unsplash, Pexels or Wikimedia Commons. The address must be the image file itself, not a page that shows it. Every external image is checked by content moderation when the page is published.

Photos from Unsplash, Pexels and Pixabay are pre-approved: those sites review everything they publish. Wikimedia Commons images are checked from a copy Webloyt fetches itself, so any file or Special:FilePath address works.

Wikimedia thumbnails only exist at certain widths (20, 40, 60, 120, 250, 330, 500, 960, 1280, 1920 and 3840 pixels, the 330px- part of the address). Pick one of those, or use the original file, and set the display size with CSS.

Small images can also be written straight into the page or stylesheet as data: images, which is handy for SVG icons and background patterns. They are checked too: SVG images get the same checks as inline SVG, and other formats are moderated like uploads.

Fonts

Fonts can come from any HTTPS address (Google Fonts, Bunny Fonts, Adobe Fonts…) or be uploaded as WOFF/WOFF2.

JavaScript libraries

Pages cannot contain JavaScript of their own: no <script> code, no event handlers. They can load these libraries with <script src=…>, which work from markup alone:

External stylesheets

Stylesheets can be uploaded to your site freely. Stylesheets from other sites are allowed only from these well-known libraries:

On cdn.jsdelivr.net this covers, for example, Bootstrap, Bootstrap Icons, Font Awesome, Bulma and Pico. For anything else, ask your AI to upload the CSS file to your site.

Content rules

Everything on Webloyt is public. Before anything is published, Webloyt checks it automatically:

• Text and images are checked for sexual content, hate, harassment, violence, self-harm and illegal activity.
• Linked websites and image hosts are checked against a blocklist of malware, phishing, pornography, gambling, violence and piracy domains.
• Every linked or loaded address is checked with Google Safe Browsing for phishing, malware and unwanted software.

Don't publish anything illegal, anything that infringes someone else's rights, or anything that pretends to be another person or organisation. Don't publish other people's personal information without their permission.

Search engines (SEO): done for you

Every Webloyt site is set up to be found by Google, Bing and other search engines, with nothing to configure.

Automatic robots.txt and sitemap.xml

• your-name.webloyt.cloud/robots.txt tells search engines they are welcome and where your sitemap is.
• your-name.webloyt.cloud/sitemap.xml lists your pages with the date each one last changed. It includes every page that can be reached by links from your homepage, and updates itself the moment you publish a change.
• Your site is also included in Webloyt's main sitemap, which is submitted to search engines.

Feedback while your AI builds

Each time a page is published, Webloyt tells your AI what search engines will miss: a missing title or description, no mobile viewport, no main heading, images without descriptions, and similar. Your AI fixes them as it goes. These are suggestions, never errors.

Built into hosting

• One address per page. /index.html redirects to /, so search engines don't see your homepage twice.
• Fast delivery. Pages and stylesheets are compressed, and unchanged files are not sent again, so returning visitors and search engines load your site quickly.
• Always fresh. Visitors always get your latest version right after you publish.
• HTTPS and security headers on every page.

Search Console

To see how Google indexes your site, add it to Google Search Console as a URL-prefix property and choose the HTML tag verification method. Ask your AI to add that <meta name="google-site-verification"> tag to your homepage.

Site icon (favicon)

Every site gets an icon for browser tabs and search results automatically: the first letter of its name on a coloured tile. To use your own, ask your AI to add one: upload an image and link it with <link rel="icon" href="icon.png">, or upload a file named favicon.ico. Yours replaces the automatic one straight away.

Keeping pages out of search results

• Don't link to it. Pages that can't be reached from your homepage are left out of your sitemap. People with the address can still open them.
• Mark it noindex. Add <meta name="robots" content="noindex"> to the page. Webloyt leaves it out of the sitemap and search engines won't list it.
• Your own robots.txt. Ask your AI to upload a robots.txt file. Webloyt serves yours instead of the default and only adds the sitemap line.

None of these make a page secret: everything on Webloyt is public to anyone who has the address.

The directory

The Webloyt directory lists sites made with Webloyt. Nothing is listed unless you ask: a site appears when its homepage carries this tag, and leaves when the tag is removed.

Ask your AI to add it when your site is ready. The content is the category: Education, Books, Business, Portfolio, Personal, Travel, Food, Science, Technology or Community; anything else is listed under Other. The directory shows your homepage's title and meta description, and its og:image if it has one, so a good description matters: ask your AI to write one that says what the site is for, in one or two sentences.

Sites under review, in the trash or marked noindex are not listed. The list refreshes within ten minutes of a change.

Agents and scripts

An AI agent or a script running on a server cannot sign in with Google in a browser. For those, Webloyt has personal tokens: a token tied to your Google account that works until you revoke it.

• Your own scripts: create a token on the tokens page and use it as the MCP server's bearer token.
• An agent asking for access: it shows you a link. Open it, sign in with Google, and approve; the agent receives its own token and continues by itself. Only approve agents you asked to use Webloyt.

A token can claim names, publish and edit pages and upload files on your account, with lower limits than a signed-in session (120 publishes an hour). It cannot delete a website. Each account can have 5 tokens; revoke any of them on the tokens page. The rules for agents, in plain text, are at webloyt.cloud/llms.txt.

Reports and review

Anyone can report a website with the “Report abuse” link at the bottom of every Webloyt site or on the report page. Reported sites may be taken offline for review. Sites that break these rules can be removed.

Questions? Contact admin@webloyt.cloud. See also our terms of use and privacy policy.