Page Agent

Page Agent is a JavaScript library that injects an AI agent directly into web pages to control user interfaces via natural language.

Page Agent functions as an in-page JavaScript script that enables AI agents to interact with web interfaces directly. It operates by performing text-based DOM manipulation, which removes the need for screenshots or multi-modal large language models. Users can integrate the agent by including a single script in their web page, allowing it to work with most mainstream LLMs, including local deployments. It is designed for developers who want to add AI capabilities to their applications without performing backend rewrites. The tool supports advanced workflows through an optional Chrome extension for multi-page tasks and an MCP server for external control.

Page Agent at a glance

Best for: Software engineer
Runs on: Web, Chrome extension, open source
Pricing: Open source under MIT License
Released: September 2025
Category: AI agent tool
Website: github.com/alibaba/page-agent

Key features

In-page JavaScript integration
Text-based DOM manipulation without screenshots
Compatibility with mainstream and local LLMs
Optional Chrome extension for multi-page tasks
MCP server support for external control

Use it for

Adding AI copilots to SaaS products
Automating complex form filling workflows
Enabling natural language accessibility for web apps
Controlling browser tabs via Chrome extension
Interfacing with agent clients via MCP

Companions

Tools that articles about Page Agent recommend using with it, each with its job.

Steel — Run the agent loop on managed remote Chromium with proxies.
Playwright — Hand-coded automation for the fixed, high-volume parts.
Ollama — Any OpenAI-compatible local model, at no cost per task.
LangChain — Compose exactly the agent your use case needs from model, tools, prompt, and middleware.
CrewAI — Gives CrewAI agents a real browser.

Alternatives to Page Agent

WebBrain — Runs locally or in the cloud with your choice of LLM and zero telemetry.
Dev Browser — Persistent browser tabs keep state across scripts so agents do not restart.
BrowserOS — Runs entirely locally and lets AI agents use real user logins to complete tasks.
Browser Use — Managed remote browsers built for protected sites at $0.02 an hour.
MyAgent — It features built-in permission controls, long-term memory, and skill self-evolution.
Webwright — Enables code-based web automation with reusable and rerunnable scripts.
Webcmd — Learns websites as agents use them to compile faster and cheaper commands.
II-Agent — It provides a fully open-source, transparent, and auditable agent foundation.
AgentGPT — It allows users to build and deploy autonomous agents in the browser.
Windows-MCP — Enables any LLM to perform native Windows tasks without complex vision models.
Goose — Runs locally on your machine with support for 70+ extensions and any LLM.
CodePilot — Consolidates multiple AI agents and local model connections in one workspace.

All Page Agent alternatives