# Page Agent review

> Page Agent is a JavaScript in-page GUI agent that lets users control web interfaces with natural language.

Devtoolwatch Board, the week of Oct 25–31, 2026. Canonical page: https://index.devtoolwatch.com/tools/alibaba-page-agent/

## TL;DR

Page Agent is a TypeScript library that adds an AI agent to any web page with one script. Page Agent is for developers building copilots, form automation or accessibility features into their own web apps. Page Agent works from the page's text-based DOM, not screenshots, and lets you bring your own LLM.

## Page Agent at a glance

- **Category:** [Browser and computer use](https://index.devtoolwatch.com/categories/browser-and-computer-use/)
- **Maintained by:** Alibaba (organization, [@alibaba](https://github.com/alibaba) on GitHub)
- **License:** MIT
- **Language:** TypeScript, JavaScript, CSS
- **Repository:** [github.com/alibaba/page-agent](https://github.com/alibaba/page-agent)
- **Website:** [alibaba.github.io/page-agent](https://alibaba.github.io/page-agent/)
- **npm package:** [page-agent](https://www.npmjs.com/package/page-agent) 1.12.4
- **Latest release:** [v1.12.4](https://github.com/alibaba/page-agent/releases/tag/v1.12.4) on GitHub, Sep 6, 2026
- **First published:** Sep 23, 2025

## What Page Agent does

- Page Agent runs as in-page JavaScript, with no browser extension, Python or headless browser required.
- Page Agent manipulates the page through the text-based DOM, without screenshots or multi-modal LLMs.
- Page Agent executes natural language instructions such as 'Click the login button' on the current page.
- Page Agent works with most mainstream LLMs, including locally deployed models.
- Page Agent offers an optional Chrome extension for multi-page tasks across browser tabs.
- Page Agent includes an MCP Server, in beta, so outside agent clients can control your browser.

### How it works

You add Page Agent to your web page with a script tag or the npm package. You then create a PageAgent with a model name, base URL and API key, and call execute with a natural language task. Page Agent reads the page's DOM as text and acts on it inside the page. An optional Chrome extension and an MCP Server extend it across tabs and to outside clients.

## How to install Page Agent

**With npm**

```
npm install page-agent
```

### Requirements

- An LLM API endpoint and API key, unless you use the free demo LLM for evaluation.

### First steps

1. Try the demo by adding the page-agent.demo.js script tag from the jsDelivr CDN to a page.
2. Install the package with npm and import PageAgent from 'page-agent'.
3. Create a PageAgent with your model, baseURL, apiKey and language settings.
4. Call agent.execute with an instruction, for example 'Click the login button'.

## Who Page Agent is for

- SaaS teams adding an AI copilot to their product without a backend rewrite.
- Developers of ERP, CRM and admin systems who want to automate form filling.
- Teams making web apps accessible through natural language and voice commands.
- Developers who want agent clients to control a browser through MCP.

### Limitations

- Page Agent is designed for client-side web enhancement, not server-side automation.
- Page Agent's demo CDN script and its free testing LLM are for technical evaluation only and are subject to terms.

## Usage data (the week of Oct 25–31, 2026)

- Board rank: #225 overall, #10 in Browser and computer use, momentum score 41
- Downloads that week: 44.2k (+35% vs 4-week average)
- New GitHub stars per week (last 12 weeks): Aug 9: 122; Aug 16: 172; Aug 23: 133; Aug 30: 112; Sep 6: 84; Sep 13: 79; Sep 20: 110; Sep 27: 127; Oct 4: 79; Oct 11: 118; Oct 18: 113; Oct 25: 71

## Page Agent alternatives

| Rank | Tool | Stars | Downloads/wk |
|---|---|---|---|
| #1 | [ego lite](https://index.devtoolwatch.com/tools/citrolabs-ego-lite/) | 21k | – |
| #2 | [Moli](https://index.devtoolwatch.com/tools/lexmount-moli/) | 21.9k | – |
| #3 | [Cua](https://index.devtoolwatch.com/tools/cua/) | 31.4k | – |
| #4 | [Browser Harness](https://index.devtoolwatch.com/tools/browser-use-browser-harness/) | 18.8k | 3.3M |
| #5 | [Jev Ultrafast](https://index.devtoolwatch.com/tools/browser-use-jev-ultrafast/) | 24k | – |
| #6 | [Stagehand](https://index.devtoolwatch.com/tools/stagehand/) | 25.9k | 3M |

## Page Agent FAQ

### Is Page Agent free and open source?

Yes. Page Agent's source code is on GitHub (alibaba/page-agent) under the MIT license, so you can use it for free.

### How do I install Page Agent?

Run `npm install page-agent`.

### Is Page Agent still maintained?

Yes. The alibaba/page-agent repository was last updated on Oct 21, 2026, and the latest release, v1.12.4 on GitHub, came out on Sep 6, 2026.

### How popular is Page Agent?

Page Agent has 29.6k GitHub stars as of Nov 1, 2026. It gained 71 new stars in the week of Oct 25–31, 2026, against a 12-week average of 110 a week. Its main package had about 44.2k downloads that week, +35% against its own 4-week average. That made it #10 among browser and computer-use agents on the Devtoolwatch Board.

### What are the best alternatives to Page Agent?

Other browser and computer-use agents on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: ego lite (#1), Moli (#2), Cua (#3), Browser Harness (#4), Jev Ultrafast (#5). The alternatives table above has their numbers.

### What is Page Agent?

Page Agent is an open-source JavaScript GUI agent from Alibaba that lives inside a web page. Page Agent lets users control the page's interface with natural language. It works on the DOM as text and uses an LLM you choose.

### How do I use Page Agent?

Install Page Agent with npm and create a PageAgent with your model, baseURL and apiKey. Then call agent.execute with a natural language task. For a quick trial, add the demo script from the jsDelivr CDN to a page.

### Does Page Agent need screenshots or a multi-modal model?

No. Page Agent uses text-based DOM manipulation. It needs no screenshots, multi-modal LLMs or special permissions.

### Which LLMs does Page Agent support?

Page Agent works with most mainstream models, including locally deployed ones. The README example uses qwen3.5-plus through DashScope. Release v1.12.3 added Kimi K3 support.

### Can Page Agent work across multiple tabs?

Yes, through the optional Chrome extension. The extension extends the agent across browser tabs for multi-page tasks. Since v1.12.0, the agent's context includes each tab's loading status.

## Sources

1. [alibaba/page-agent on GitHub](https://github.com/alibaba/page-agent)
2. [Page Agent README](https://github.com/alibaba/page-agent/blob/main/README.md)
3. [page-agent on npm](https://www.npmjs.com/package/page-agent)
4. [Page Agent website](https://alibaba.github.io/page-agent/)
5. [Release v1.12.4 (Sep 6, 2026)](https://github.com/alibaba/page-agent/releases/tag/v1.12.4)
6. [Release v1.12.3 (Sep 5, 2026)](https://github.com/alibaba/page-agent/releases/tag/v1.12.3)
7. [Release v1.12.2 (Jul 16, 2026)](https://github.com/alibaba/page-agent/releases/tag/v1.12.2)
8. [Hacker News: Show HN: PageAgent, A GUI agent that lives inside your web app](https://news.ycombinator.com/item?id=47264138)
9. [Hacker News: Weex - A framework for building Mobile cross-platform UI](https://news.ycombinator.com/item?id=11927949)
10. [Devtoolwatch Board, the week of Oct 25–31, 2026](https://index.devtoolwatch.com/)