teramotodaiki

teramotodaiki / human-browser

Public

Logged-in Chrome automation for agents

32
1
100% credibility
Found Feb 17, 2026 at 30 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

Human Browser is a local automation tool that connects a Chrome extension to a background service, allowing simple commands to control web pages by capturing interactive elements and performing human-like actions such as clicking, typing, scrolling, and screenshotting.

How It Works

1
🕵️ Discover the tool

You hear about Human Browser, a handy way to make your web browser follow simple instructions like a helpful assistant.

2
🛠️ Set it up quickly

You download the pieces, add a tiny helper to your Chrome browser, and start a background service on your computer.

3
🔗 Connect everything

You share a private code between the browser helper and service, and watch the icon turn green as they link up securely.

4
🌐 Open your webpage

You fire up Chrome, head to any website, and pick the tab you want your assistant to work with.

5
📸 Snap a page picture

You ask for a quick snapshot, and get a friendly list of all buttons, links, boxes, and interactive spots on the page.

6
🎮 Give easy instructions

You tell it to click a button, type in a form, scroll down, or hover over something, feeling like directing a robot friend.

🎉 Enjoy the results

Your tasks finish automatically with screenshots, grabbed text, or completed actions, saving you tons of clicking time.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 30 to 32 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is human-browser?

Human-browser lets you automate real Chrome tabs with persistent logins via a secure local daemon and Chrome extension bridge. Developers run a Node.js CLI to snapshot page structure into accessible trees (roles, names, selectors), then click, fill forms, hover, scroll, or wait using refs like `@e5` or CSS selectors—all while keeping Chrome logged-in accounts active. It solves agentic browser control without CAPTCHA hurdles or headless auth loss, using WebSocket for low-latency commands.

Why is it gaining traction?

Unlike Puppeteer or Playwright, which struggle with logged-in sessions and human verification, this runs on your everyday Chrome with stay-logged-in cookies intact. The snapshot refs make interactions reliable across page changes, and CLI flags like `--interactive` or `--compact` deliver precise DOM intel fast. Devs hook it for GitHub logged-in scripts or Chrome logged-in devices without rebuilding sessions.

Who should use this?

AI agent builders scripting logged-in flows on GitHub, Twitter, or banking sites. Frontend testers needing reproducible clicks/fills in real browsers. Ops folks monitoring network/console in persistent Chrome tabs without headless friction.

Verdict

With 30 stars and 1.0% credibility, it's early-stage—light docs, no tests visible—but solid for niche logged-in Chrome automation. Try the CLI init/daemon flow if headless fails you; rotate tokens for security.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.