Two Free Open‑Source Skills That Seamlessly Replace Playwright

When using agents for browser automation, the author previously relied on Playwright or Chrome DevTools. Simple tasks work, but real‑world scenarios—such as QR‑code login, session persistence, multi‑account handling, and captcha challenges—expose Playwright’s lack of specialized optimizations.

Introducing BrowserAct

BrowserAct is an open‑source project designed for AI agents. It offers a CLI ( browser‑act ) for real‑time browser control and a skill‑forge ( browser‑act‑skill‑forge ) that packages website capabilities into reusable skills.

Core Features

Stealth Browser : a fingerprint‑masked browser that bypasses anti‑scraping mechanisms, includes dynamic proxy rotation, and integrates automatic captcha solving.

Dynamic Proxy : automatically rotates IPs by region for large‑scale data collection or geo‑restriction bypass.

Session Isolation : each task runs in an independent browser environment, preventing cross‑contamination of cookies, sessions, and proxies.

Human‑Agent Collaboration : a remote‑assist layer generates a link for a human to complete required verification steps (e.g., scanning a QR code) before the agent resumes.

Browser Modes

Stealth : creates a fresh anti‑detection instance with unique fingerprint and proxy; ideal for breaking anti‑scraping defenses and parallel multi‑account crawling.

Chrome (Login‑State Reuse) : launches Chrome with existing cookies, suitable for interacting with already‑logged‑in services.

Chrome‑Direct : connects via CDP to the currently running Chrome, enabling quick debugging and seamless hand‑off between human and agent.

Skill‑Forge Capabilities

browser‑act‑skill‑forge discovers a website’s API endpoints and request patterns, then automatically generates a SKILL.md and a Python package. Once created, the skill can be reused for batch or scheduled tasks, dramatically reducing manual effort.

请你读取这个链接，帮我安装里面的 Skill，并测试一下能否正常运行：
https://github.com/browser-act/skills

Practical Examples

Using the CLI, an agent can simultaneously perform tasks such as checking recent GitHub issues, fetching the latest posts from a Xiaohongshu user, listing trending open‑source projects, and retrieving recent YouTube channel content.

用 browser‑act 同时并行完成以下任务：
① 查看开源项目最近的 issues 和 PR： https://github.com/Wechat-ggGitHub/Awesome-GitHub-Repo
② 搜索小红书博主最近的帖子： https://www.xiaohongshu.com/user/profile/548251dce779893bcf3f77bc
③ 查看热门开源项目： https://github.com/trending
④ 获取 Lex Fridman 频道最近 10 期内容： https://www.youtube.com/@lexfridman

Skill Ecosystem

BrowserAct ships with 31 ready‑to‑use skills covering five major domains:

E‑commerce (8) : ASIN lookup, bestseller search, Buy Box monitoring, competitor analysis, listing comparison, product details, product search, review extraction.

Lead Generation (7) : merchant contacts, GitHub contributor lookup, Google Maps business search and reviews, industry radar, social media discovery.

Search Research (4) : Google Image search, News, web research assistant, web scraping.

Social Listening (3) : Reddit competitor analysis, WeChat public account search, Zhihu search.

Video Platforms (9) : YouTube search, channel analysis, comment extraction, subtitle extraction and batch processing, KOL discovery, video details, YouTube API.

All skills have been validated in real scenarios and can be deployed instantly.

Benefits

High efficiency and cost savings by reusing learned automation paths.

Automatic stripping of 90% irrelevant HTML (ads, trackers, framework noise), delivering clean content to LLMs.

Robust multi‑task handling with isolated sessions for each account.

Overall, BrowserAct provides a comprehensive, open‑source alternative to Playwright, especially for AI‑driven agents that need stealth, proxy management, and reusable automation skills.