Skip to content

Extract the readable content of web pages

POST/v1/tools/web.extract/callsNeeds an API key

Returns the readable content of one web page, as Markdown and as plain text, with its title and description. Use it to give an agent a page to read.

Tool web.extract on Open web and SEO. Contract version 1, released 2026-10-27.

Over REST, send the input in the body. Over MCP, pass the same input to call_tool.

Both return the full response: the call's ID, status and cost, with this tool's result in output.

Input

urlstringRequired

Address of the public page to extract, starting with http:// or https://.

Format uri.

Matches ^https?://[^\s/?#]+\S*$.

Output

A successful call returns these blocks in its output. A field not marked required can be null.

page

One object. Can be null.

contentstringRequired

The whole page as Markdown.

textstring

The readable text of the page, without markup.

titlestring
descriptionstring

The description the page declares for itself. Absent when it declares none.

urlstring

Address of the page that was read.