pasky

pasky / chrome-cdp-skill

Public

Give your AI agent access to your live Chrome session — works out of the box, connects to tabs you already have open

256
10
100% credibility
Found Mar 13, 2026 at 150 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

A skill that enables AI agents to view and interact directly with tabs in a user's running Chrome browser instance.

How It Works

1
🔍 Discover the Tool

You hear about a handy way to let your AI helper peek into and play with your open web browser tabs without starting over.

2
🖥️ Flip the Switch in Your Browser

In your everyday web browser, go to a special page and turn on one simple sharing option to allow access.

3
Connect to Your AI Helper
⭐
For Your Favorite AI Buddy

Simply tell it to pick up this new skill, and it's ready in moments.

📁
For Other Smart Tools

Copy a folder into your AI's working area, and it works right away.

4
📋 See Your Open Tabs

Run a quick check to list all the tabs you have open, so your AI knows what's available.

5
📸 Capture and Interact

Snap pictures of pages, read content, click buttons, or type just like you're doing it yourself—but your AI does the work.

✅ Seamless Control Achieved

Now your AI can jump into your real browser sessions, staying logged in and keeping everything just as you left it.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 150 to 256 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 chrome-cdp-skill?

This JavaScript tool gives your AI agent access to your live Chrome session—the tabs you already have open, logged-in accounts like GitHub, and current page states—without launching a new browser or re-logging in. Built on Chrome DevTools Protocol and Node.js 22+, it offers simple CLI commands to list tabs, screenshot pages, grab HTML, evaluate JS, click elements, type text, and more. Enable remote debugging in Chrome with one toggle, and your agent can interact with real workflows instantly.

Why is it gaining traction?

Unlike Puppeteer or chrome-devtools-mcp, which reconnect per command and struggle with many tabs, this holds persistent sessions per tab—approve debugging once, then run reliably even with 100+ tabs open. Developers love the out-of-the-box setup: copy a folder for agents like Cursor or Claude Code, or install via pi, and start giving agents access to private GitHub repos or internal tools through your existing session. The CLI's targeted actions, like scoped HTML pulls or "load more" loops, deliver quick wins without framework overhead.

Who should use this?

AI agent builders integrating browser control into tools like Claude Code or Amp, especially for tasks needing logged-in access such as giving GitHub repo read access to private projects or automating GitHub collaborator permissions. Frontend devs testing interactive UIs in live sessions, or Road 96-style game devs scripting agent behaviors via Chrome tabs. Skip if you need cross-browser support.

Verdict

Promising for agent-driven browser hacks, with solid docs and user-friendly CLI, but at 47 stars and 1.0% credibility score, it's early-stage—test thoroughly before production. Worth a spin if you're giving agents access to classification apps or soil erosion sims in tabs.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.