BrowserSkill

BrowserSkill is an open-source tool from Tencent that lets AI agents use your real, logged-in browser — with your existing sessions and accounts — without interrupting your work. Released under the MIT license, it has collected more than 9,300 stars on GitHub and was actively developed through 2026.

Overview

Most browser automation gives an agent a fresh, logged-out browser, which fails the moment a task needs your authenticated session. BrowserSkill takes the opposite approach: a command-line interface plus a browser extension connect shell-capable AI agents to the browser you actually use, so agents can operate sites you are already signed into. Because the agent drives a separate execution path, it can work in the background while you keep browsing, and any agent that can run shell commands can use it — no per-framework integration needed.

Key Features

• Gives AI agents access to your real browser with your logged-in sessions
• CLI-first design that works with any shell-capable AI agent
• Browser extension bridges agent commands into live pages
• Designed to avoid interrupting your own browsing while the agent works
• Useful for tasks behind logins that fresh-browser automation cannot reach
• Open-source and self-hosted, released by Tencent under the MIT license

💳 Pricing

BrowserSkill is free and open source under the MIT license. You bring your own AI agent and any model subscriptions it uses.

Because it speaks to agents through a plain command-line interface, BrowserSkill slots into existing agent setups without a dedicated plug-in for each framework — if the agent can run a shell command, it can drive the browser. That design makes it equally useful for one-off personal automations and for evaluating how agents behave on real, stateful websites.

Who Is It For

It is for agent builders and power users whose automations keep hitting login walls — internal tools, dashboards and web apps where the useful work only exists inside an authenticated session.