Browser automation

When a page needs a real browser — JavaScript rendering, modern frameworks, login-aware pages — Han AI drives a headless Chromium via Playwright on the VPS.

What it does

Loads a page in a real browser, waits for it to render, and returns either the visible page content or a clean structured scrape.

Field	Value
Schema names	`browse`, `scrape_clean`
Powered by	Playwright plus Chromium
Browser path	`/var/hanai/playwright-browsers`
API key required	No

When Han AI uses it

A static page fetch returned mostly empty markup because the page is a single-page app.
A site requires interaction (a click, a scroll, an accept) before its content is visible.
A scrape needs to be clean of nav chrome, ads, and pop-overs.

Examples

“Pull the live listings from that broker’s portal page.”
“Get the current room availability from the property management site.”
“Open this page, accept the cookie banner, and read me the article body.”

Limits

Slower than plain page fetch — seconds, not milliseconds.
Heavily-protected sites (Cloudflare, hCaptcha, fingerprinting) may still block headless Chromium.
Long-running interactive flows should be modeled as recurring tasks rather than packed into a single live turn.

Why this stack

Playwright with Chromium is the industry-standard open-source browser automation stack. Running it on your VPS means dynamic pages can be read without paying a hosted scraping service, and without your queries being logged by a third party.