Description
Page Agent is a GUI agent that lives inside your web page. You add one script tag to a website, and users can then type natural language commands such as fill the expense report for Friday's trip, and the agent fills in the form fields and carries out the steps for them. It runs entirely in the browser as a front-end library, so you do not need a Python environment, a headless browser or a server. It works with many language models, including OpenAI, Claude, DeepSeek, Qwen, Gemini and Grok, and it can run offline with Ollama or LM Studio. The agent asks for confirmation before acting, through a built-in collaborative panel, so it is not blind automation. It can also handle multi-page tasks and be controlled from outside the page.…
Highlights
- One script tag to add an AI agent to a page
- Natural language commands for web forms
- Runs in the browser with no server
- Works with many models, including offline Ollama
- Human-in-the-loop confirmation
Ratings & Reviews
No ratings yet
Be the first to rate PageAgent.
People also visit
PageAgent FAQ
What is Page Agent?
A GUI agent in JavaScript that lets users control a web page with natural language commands.
How do I add it to my site?
Add the page-agent script to your page, or install the package from npm and create a PageAgent with a model and API key.
Which models can it use?
OpenAI, Claude, DeepSeek, Qwen, Gemini, Grok, Kimi, GLM and more, or fully offline with Ollama or LM Studio.
Is it free?
Yes. It is MIT licensed and the source is on GitHub. You pay your model provider for any API use.
What are the best PageAgent alternatives?
Popular alternatives to PageAgent include Agent Page, Web-Use and Agent Checker. See all PageAgent alternatives






