Page Agent is a JavaScript-based in-page GUI agent developed by Alibaba that enables users to control web interfaces through natural language commands. Written in TypeScript, the project provides a lightweight alternative to traditional browser automation by operating entirely within a webpage without requiring browser extensions, Python, or headless browsers. The core functionality centers on text-based DOM manipulation, eliminating the need for screenshots or multi-modal large language models while allowing developers to bring their own LLM implementations.
The repository has demonstrated steady growth, with GitGenius tracking an increase of 9 stargazers between July 4, 2026 and the present, bringing the total to 23,062 stars. The project maintains active development with a median issue and pull request response latency of 0 hours and a mean latency of 4.3 hours across 190 tracked items. The most active contributor, gaomeng1900, has logged 574 events, while AburronKing and liujianing971218 have each contributed 12 events. Bug reports and enhancement requests dominate the issue tracker with 55 and 50 items respectively, with 17 items marked as high priority.
Page Agent is classified across multiple monitoring and analytics categories including front-end monitoring, performance monitoring, real user monitoring, web performance, user experience, error tracking, browser agents, data collection, JavaScript errors, and page analytics. The project maintains connections with other significant repositories through overlapping contributors, including nuxt/nuxt, langgenius/dify, and openclaw/openclaw, indicating its integration within a broader ecosystem of web development and AI agent tools.
The primary use cases outlined in the repository include deploying AI copilots within SaaS products without backend rewrites, automating smart form filling for complex workflows in ERP and CRM systems, improving accessibility through natural language interfaces, extending agent capabilities across multiple browser tabs via a Chrome extension, and supporting Model Context Protocol servers for external agent control. The project offers multiple integration paths, from a single-line CDN integration using a free demo LLM to full NPM installation for more programmatic usage.
Page Agent distinguishes itself through its client-side design philosophy, explicitly positioning itself for web enhancement rather than server-side automation. The project provides optional Chrome extension support for multi-page tasks and a beta MCP Server implementation. Integration is designed to be straightforward, with no need for special permissions or complex setup procedures. The repository maintains clear contribution guidelines and explicitly states that contributions generated entirely by bots or AI without substantial human involvement will not be accepted, emphasizing the importance of human-driven development.
The project is licensed under the MIT License and acknowledges its foundation on the browser-use project. Documentation is comprehensive, with a dedicated homepage at alibaba.github.io/page-agent providing demos, guides, and developer resources. The repository actively encourages community contributions and maintains a section for community-built projects using Page Agent.
