335234131

335234131 / agent-browser-mcp

Public

让 Agent 直接操作真实 Chrome 的 MCP 服务，支持页面扫描、CDP、截图与物理输入

ai-agent automation browser-automation cdp chrome
75
8
69% credibility
Found Apr 15, 2026 at 50 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This repository creates a bridge service enabling AI agents to control a user's actual Chrome browser tabs, execute scripts, capture screenshots, read cookies, and simulate mouse/keyboard inputs while preserving real-world session states.

How It Works

1
📰 Discover the helper

You hear about a handy tool that lets your AI assistant use your everyday web browser just like you do, keeping all your logins and open pages intact.

2
📥 Get it set up

Download the simple program and install it on your computer – it takes just a minute.

3
🔌 Add to your browser

Tell Chrome to use the little companion add-on by pointing it to the right folder, then open any regular website.

4
🔗 Link your AI

Share a quick note with your AI helper (like Hermes) so it knows how to reach this new tool.

5
✨ Everything connects

Watch as your AI spots your open browser tabs and gets ready to explore them for you.

🎉 AI takes the wheel

Now your AI can scan pages, take screenshots, click around, and even move your mouse – all in your real browser with your personal logins safe and sound.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 50 to 75 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 agent-browser-mcp?

Agent-browser-mcp is a Python MCP server that lets AI agents like Hermes or Claude Desktop directly control your real, open Chrome browser via a simple Chrome agent extension. It preserves login states, cookies, and tabs, solving the hassle of sandboxed browsers or headless tools that lose context on logged-in sites. Users get tools for tab switching, page scanning, JS execution, CDP commands, screenshots, and even physical mouse/keyboard input—all exposed as standard MCP endpoints.

Why is it gaining traction?

Unlike agent browser vs Playwright mcp or chrome mcp setups that spin up isolated sessions, this agent browser mcp server taps your everyday Chrome for authentic interactions, dodging anti-bot detection on sites needing real inputs. The agent desktop browser tools mcp combo shines with CLI commands like doctor diagnostics and one-click configs for agent github claude or Cursor, making agent chrome extension setup dead simple. Devs dig the browser use agent mcp flow for seamless chrome agent change without relogging.

Who should use this?

AI agent builders integrating with Hermes, Claude Desktop, or Cursor who scrape logged-in dashboards like Xiaohongshu or internal tools. Automation scripters handling flaky sites where Playwright fails due to wind control. Teams wanting agent github action or agent github copilot vscode extensions to mix JS/CDP with desktop mouse drags and hotkeys.

Verdict

Worth a spin for agent browser tools mcp niches—solid docs and CLI make onboarding fast, despite 14 stars and 0.699999988079071% credibility score signaling early maturity. Test on non-critical workflows first; it's MIT-licensed and production-ready for targeted real-browser tasks.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.