Your agent asks. The web answers in 148 ms.

Two calls. Discover finds the pages. Extract reads a page as clean Markdown or JSON.

25 free credits each month. No credit card.

$extraktor extract https://nodejs.org/en/about/previous-releases148 ms

| Node.js | Codename | Last updated | Status | || --- | --- | --- | --- | --- || [v 26](https://nodejs.org/en/download/archive/v26.11.1) | - | Oct 07, 2026 | Current | || [v 25](https://nodejs.org/en/download/archive/v25.9.0) | - | Mar 31, 2026 | EOL | || [v 24](https://nodejs.org/en/download/archive/v24.21.0) | Krypton | Sep 07, 2026 | LTS | |

previous-releases.md · Markdown, metadata and schema.org data

Every call on this page is real.

We recorded these calls on 2026-10-09 and 2026-10-10. Select a call to see what your agent gets.

POST/v1/search { "query": "cloudflare workers cpu time limit" }2,752 ms

Send a query. Get the top Google results of the moment: titles, links and snippets. Search does not load the pages, so it is fast.

[  { "title": "Custom limits · Cloudflare for Platforms docs", … },  { "title": "Increase CPU limit in new Workers pricing model", … },  { "title": "[🐛 Bug]: Worker exceeded CPU limit. · Issue #700", … }]

200 · organic_results.json

Also: your own JSON schema, site links, screenshots and PDFs. See the docs

Use it from your stack.

Coding agents

Claude Code, Codex, Cursor, VS Code, Gemini CLI and Windsurf.

npx -y @extraktor/cli mcp add

MCP server

For the Claude API, the OpenAI Responses API and the AI SDK.

https://extraktor.app/mcp

REST API

Any language. No SDK. One OpenAPI file.

https://extraktor.app/openapi.json

One credit for each call that works.

One credit for each successful search or extraction, in the web app, the API, MCP and the CLI.

plan	credits / month	for	USD / month	Action
Free	25	Trying it out	$0	Start
Starter	1,000	Regular use	$19	Choose
Pro	3,000	Heavy use	$39	Choose

Failed requests and HTTP error pages do not use a credit.

100 extractions and 100 searches per minute. No rollover. Cancel at any time.