Tencent / BrowserSkill

项目摘要

BrowserSkill is a CLI and browser extension that connects AI agents to a user's real, logged-in Chrome or Edge browser, enabling agents to read pages, fill forms, capture screenshots, and debug websites through a separate Agent Window without interrupting the user's workflow.

为什么值得关注

It solves the practical problem of giving AI agents access to authenticated browser sessions while keeping the user's own browsing undisturbed, supporting multiple popular AI agents (Cursor, Claude Code, Codex, etc.) with active development and frequent releases.

优势

• Works with existing logged-in browser sessions without requiring re-authentication
• Separate Agent Window prevents interruption of user's own browsing
• Broad AI agent compatibility (Cursor, Claude Code, Codex, OpenClaw, and others)
• Strong debugging capabilities with evidence capture, request inspection, and JSON export
• Active development with 13 releases in 6 months and multi-platform support

局限性

• No automated tests despite having CI
• No Docker support for containerized environments
• No examples directory to help new users get started
• Limited to Chromium-based browsers (Chrome, Edge)

使用场景

• Automating web workflows and form filling using existing login sessions while continuing to work
• Debugging websites by connecting AI agents to network requests, console logs, and page changes
• Capturing full-page screenshots and evidence for documentation or bug reports
• Investigating failing API requests and performance issues with AI-assisted analysis