Extraction workflow
How Webpage-to-Text Extraction Works
WEB/TEXT validates one public HTTPS URL, fetches bounded HTML through a Cloudflare Worker, and gives only temporary script-blocked HTML to Browser Run for text or Markdown conversion.
Start with one public HTTPS URL
Submit one publicly reachable HTTPS webpage. WEB/TEXT does not accept browser cookies, account credentials, authorization headers, private files, or local-network addresses.
Keeping the input boundary narrow makes the result easier to explain and reduces the risk of treating a private page as public source material.
Validate redirects and fetch public HTML
A Cloudflare Worker validates the initial URL and DNS answers, then fetches the page through the public internet. It handles redirects manually and validates each redirect before following it, with a fixed hop and HTML size limit.
The fetch does not include application-supplied cookies, Authorization, custom headers, or secrets. This route handles one user-requested public page, not site crawling or a credentialed browsing session.
Convert isolated HTML with Cloudflare Browser Run
Cloudflare Browser Run receives temporary HTML, not the submitted target URL, browser cookie, Authorization, custom header, or secret. JavaScript is disabled, and scripts and every subresource type are blocked before the HTML is converted to Markdown.
Static and server-produced HTML can be readable, but JavaScript-only pages may fail. WEB/TEXT reports a clear error instead of enabling scripts, retrying around controls, or presenting an empty page shell as complete content.
Choose readable text or structured Markdown
Plain text is for reading, quoting, notes, and systems that do not need document formatting. Markdown preserves headings, lists, links, tables, and code when structure carries meaning.
The output stays traceable to the source. It does not add AI-written claims or silently summarize the page. When readable content cannot be produced, WEB/TEXT returns a clear error.
Keep source context with the current request
Readable output is more useful when you can verify where it came from. The result surface shows the source domain, path, word count, and reading time next to the extracted document for the current request.
WEB/TEXT does not persist extracted page bodies or create a public result page. See the privacy page for the separate target-site fetch and temporary Browser Run HTML boundaries.