Chrome-GPT is an open-source autonomous agent that controls a real Chrome browser on your desktop. Built with LangChain and AutoGPT-style planning on GPT-4, it takes a goal in natural language โ research a topic, compare products, fill in information โ and then clicks, types and navigates through web pages step by step to carry it out, narrating its plan as it goes. The project has around 1,700 stars; its code was last updated in 2023.
Chrome-GPT was one of the early wave of AutoGPT experiments applied to the browser. The agent observes the current page, decides the next action (clicking a link, entering text, scrolling), executes it through browser automation, and loops until the goal is complete or it gets stuck. It is a compact, readable codebase that shows the full agent loop end to end, which made it a popular starting point for people building their own web agents.
Free. Chrome-GPT is open source under the GPL-3.0 license; you only pay for your own OpenAI API usage.
Developers learning how autonomous web agents work, tinkerers automating repetitive browsing tasks, and researchers comparing early AutoGPT-style agent implementations.
Because it drives a real Chrome browser rather than a stripped-down sandbox, Chrome-GPT can interact with JavaScript-heavy websites, and its step-by-step action log doubles as a debugging trace whenever the agent takes a wrong turn.