Overview

The AI agent is the core of ScrapeSpace. It controls a real cloud browser using screenshots and DOM analysis — the same way a human would browse, but faster and more precise.

What the agent can do

Navigation

Navigate to any URL, go back, reload, and switch between browser tabs
Switch into iframes to interact with embedded content
Wait for elements or navigation to complete before continuing

Interaction

Click, double-click, and hover over any element
Type into inputs, search boxes, and forms
Select options from dropdowns
Press keyboard keys and shortcuts (Enter, Tab, Escape, Ctrl+A, etc.)
Scroll pages, or scroll directly to a specific element
Expand collapsible sections and accordions
Low-level mouse and keyboard control for elements that resist normal interaction

Data extraction

Extract text or HTML attributes (href, src, etc.) from individual elements
Extract structured data from repeated lists (tables, cards, search results)
Run JavaScript directly on the page for complex or bulk extraction
Make HTTP requests to a site's API endpoints
Handle pagination — next buttons, infinite scroll, page numbers

Authentication & security

Log in to websites using your stored credentials
Generate one-time passwords (TOTP) for two-factor authentication
Solve CAPTCHAs automatically — Cloudflare Turnstile, hCaptcha, reCAPTCHA v2/v3, FunCaptcha, GeeTest v3/v4, Amazon WAF, and MTCaptcha
Save and load cookies to maintain sessions

What the agent cannot do

The agent controls a real browser, so it can do anything you could do manually. However, it will reject prompts that have nothing to do with browser automation:

Math problems, trivia, or general knowledge questions
Writing essays, emails, or other text
Anything that doesn't involve a website

Live activity

While the agent is working, you can watch its progress in real-time via the activity log. Each action is logged with what the agent did and why, so you can understand its decision-making process.