Repolode
Submit
Tencent

Tencent/BrowserSkill

A CLI and browser extension that let AI agents work inside your logged-in Chrome or Edge while you keep using the browser.

View on GitHub
Stars
9,149
Forks
658
Open issues
104
Last commit
4 days ago

What is it?

BrowserSkill connects a shell-capable AI agent to Chrome or Microsoft Edge, using the accounts you are already signed into. Tasks such as reading pages, filling forms, navigating multi-step websites, capturing screenshots, or investigating a failing request run in a separate, visible Agent Window. An existing tab can be explicitly borrowed for a task and returned when the task ends.

The project is made of a `bsk` CLI with a background daemon, a browser extension, and an agent skill that teaches the agent how to call the CLI. It works with agents such as Cursor, Claude Code, Codex, and OpenClaw, and includes a dedicated plugin for DeepSeek Harness with native browser tools. The CLI runs on macOS, Linux, and Windows, and the extension requires Chromium 125 or later.

Beyond basic automation, it offers website debugging that correlates actions with requests, response bodies, console messages, and performance data, plus structured table extraction, full-page screenshots, and optional task video recording. Users can choose a specific browser profile or pair a server-side agent with a browser on their own computer over authenticated WSS.

Key features

  • 01Runs agent tasks in a separate visible window that shares your current login state.
  • 02Reads page text, clicks, types, manages tabs, and captures viewport or full-page screenshots.
  • 03Correlates browser actions with requests, responses, console output, and performance metrics for debugging.
  • 04Exports loaded tables and lists as JSON or CSV.
  • 05Lets you name browser instances and bind a session to a specific profile.
  • 06Supports pairing a server-hosted agent with a browser on your computer.

Use cases

  • Having an agent read internal documents or search an intranet site that requires your login.
  • Letting an agent fill out and submit multi-step web forms while you keep working.
  • Investigating why a form save or API call fails on a site you are authorized to debug.
  • Exporting table data from a web page into CSV for analysis.

Topics

#browser-automation#ai-agent#chrome-extension#cli#web-debugging#browser-use#agent#dsh-plugin

FAQ

What is BrowserSkill?
A CLI and browser extension that let AI agents work inside your logged-in Chrome or Edge while you keep using the browser. BrowserSkill connects a shell-capable AI agent to Chrome or Microsoft Edge, using the accounts you are already signed into. Tasks such as reading pages, filling forms, navigating multi-step websites, capturing screenshots, or investigating a failing request run in a separate, visible Agent Window. An existing tab can be explicitly borrowed for a task and returned when the task ends.
Is BrowserSkill free to use? What license does it have?
BrowserSkill is open source under the MIT license. Check the license terms before using it in commercial products.
Is BrowserSkill still maintained?
Actively maintained. The last commit to BrowserSkill was on Oct 15, 2026.
What can I use BrowserSkill for?
Having an agent read internal documents or search an intranet site that requires your login. Letting an agent fill out and submit multi-step web forms while you keep working. Investigating why a form save or API call fails on a site you are authorized to debug. Exporting table data from a web page into CSV for analysis.
Where can I get BrowserSkill?
The source code and installation instructions are on GitHub: https://github.com/Tencent/BrowserSkill

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects