BrowserMan

BrowserMan is a browser automation tool for AI agents that works with a signed-in Chrome session. It is designed for everyday web work such as gathering information, organizing leads, and supporting operations, while keeping the browser, cookies, and credentials on the user’s device.

The service combines general browser control with prebuilt actions for repeated workflows. It says agents can work on any website using core browser tools such as navigate, read, click, and type, and that it also offers 13+ prebuilt actions for major platforms including X, LinkedIn, Reddit, Medium, Xiaohongshu, Weibo, Bilibili, Zhihu, Douyin, Substack, Product Hunt, YouTube, and Hacker News. BrowserMan describes these actions as a way to use fewer tokens and avoid repeated page exploration, and it notes that each platform action is a typed call rather than a fragile chain.

Access can be scoped per agent or vendor, including which accounts, sites, and actions are available. The page says every click, navigation, and post is logged and attributed to the agent that made it, and access can be revoked from the dashboard in one click. For high-risk actions such as posts, purchases, deletions, or sends, the BrowserMan skill asks for confirmation in chat before proceeding.

BrowserMan is installed as a Chrome extension from the Chrome Web Store. After installation, the browser is paired from the extension side panel, and the agent is connected through MCP or an HTTP API. The page names Claude, Cursor, ChatGPT, n8n, and other MCP clients as examples. It says the tool uses a thin hosted relay at browserman.run between the agent and the browser, and that page content is returned only when requested. BrowserMan is free to use, and the page says delegated tokens can be issued and revoked at any time. It also says sign-in is required, with Google or GitHub available.