Page Agent Review 2026: Control Any Website With Natural Language

Browser automation usually means writing selectors and waiting on fragile XPath. Page Agent flips that: you describe what you want in natural language, and an agent plans and executes the steps on the page.

What is Page Agent?

Page Agent is an open-source, JavaScript in-page GUI agent from Alibaba. It runs as both a Chrome extension and an npm package, letting you control any web interface with natural language — click buttons, fill forms, navigate, and extract structured data.

Key features

• Natural-language control of web UIs — no selectors to maintain
• Chrome extension + npm package — use it in the browser or in code
• Open source (MIT), backed by Alibaba
• Plan-and-act agent for multi-step web tasks
• Structured data extraction from pages

How it helps

Instead of scripting a flow, you prompt it: “log into the dashboard, open last month’s report, and copy the total into this sheet.” The agent breaks that into steps and executes them, which is far faster to prototype than writing Playwright from scratch.

Pros and cons

Pros: natural-language control, dual extension/package form, open source with a strong backer, good for multi-step tasks.

Cons: works best on JavaScript-heavy pages; browser automation can be brittle on highly dynamic sites; the project is recent and still evolving.

Pricing

Free and open source (MIT).

FAQ

Is it the same as Browser Use? Similar idea, different implementation — Page Agent is Alibaba’s in-page agent with both an extension and an npm package.

Can it extract data? Yes, it can pull structured data out of pages as part of a task.