WebBrain is an open-source AI browser agent that runs inside Chrome and Firefox, letting an AI model operate your browser on your behalf. The project has gathered around 1,500 stars on GitHub and is under very active development in 2026, with a focus on running modern open models locally for privacy and offline use.
WebBrain ships as a browser extension built from a monorepo, pairing the extension front end with agent logic that can see page structure, plan multi-step tasks, and act through clicks, typing, and navigation. Its topic set tells the story: it is designed around locally runnable model families such as Gemma, Qwen, and MiniMax, so browsing assistance does not have to mean sending every page to a third-party cloud. That makes it usable as an offline AI assistant for tasks like reading and summarizing pages, filling repetitive forms, extracting information across tabs, and automating research workflows. Because it integrates through standard Chrome and Firefox extension APIs, including DevTools-level inspection, the agent can work with the pages you actually use rather than a sandboxed demo environment. The project is open source and community-driven, and it suits users who want browser automation with the model under their own control.
Free. WebBrain is an open-source project distributed at no cost; you supply the local or remote model you want it to run on.
Privacy-conscious users and developers who want a browser agent that can run on local open models instead of depending on a proprietary cloud assistant.