WebOperator

An AI browser agent for your existing Chrome sessions, with local models through Ollama or MLX, scheduled tasks and an MCP bridge.

WebOperator is an AI browser agent for automating tasks in the Chrome tabs you already use. It retains your existing logins and sessions, and can use a local model or a remote provider.

Browser tasks: Give it a plain-language goal to interact with pages, switch tabs, take screenshots and extract structured text; it checks the page after each action.
Visible task history: Review plans, actions and answers in the side panel, with saved traces and checkpoints for inspecting completed runs.
Scheduled automation: Run recurring tasks with the side panel closed; runs pause for your attention when they encounter a login or verification challenge.
MCP access: A local bridge lets Hermes, OpenClaw and other MCP clients use your active browser tab as a tool.

Ollama and MLX provide local inference, so page observations can stay on your machine. Choosing a remote provider sends that provider page text and screenshots. The model must support tool calling; vision and reasoning capabilities are optional.

WebOperator is open source under the MIT license. Chrome and Brave are tested browsers, while Windows and other Chromium browsers are untested. The MCP bridge runs on macOS and Linux. The extension isn't available through the Chrome Web Store.

The agent can only use content visible to it in the browser. Bot detection can block actions, dynamic or canvas-heavy pages can disrupt interactions, and long tasks can drift despite checkpoints and context compression.