惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

云风的 BLOG
云风的 BLOG
The GitHub Blog
The GitHub Blog
Y
Y Combinator Blog
博客园 - 三生石上(FineUI控件)
T
The Blog of Author Tim Ferriss
宝玉的分享
宝玉的分享
Hugging Face - Blog
Hugging Face - Blog
WordPress大学
WordPress大学
V
Visual Studio Blog
小众软件
小众软件
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
MongoDB | Blog
MongoDB | Blog
V
V2EX
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
博客园 - 【当耐特】
Microsoft Azure Blog
Microsoft Azure Blog
The Cloudflare Blog
H
Hackread – Cybersecurity News, Data Breaches, AI and More
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
Engineering at Meta
Engineering at Meta
L
LangChain Blog
Martin Fowler
Martin Fowler
GbyAI
GbyAI
博客园 - 司徒正美

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
I Built a VS Code Extension That Lets You Control GitHub ...
Atul Srivast · 2026-05-11 · via DEV Community

The "Aha" Moment That Started Everything

Three months ago, I gave GitHub Copilot agent mode a massive refactoring task — rewrite an entire auth module across 12 files.

I hit Enter. Walked to the kitchen. Made chai. Came back 40 minutes later.

The agent had stopped after 2 minutes and 38 seconds.

It was waiting for me to type "Yes, proceed."

38 minutes. Gone. Just like that.

I stared at my screen and thought: "This is the most powerful coding tool ever built... and it's useless the moment I stand up."

That's when I started building Copilot Remote Control.


What Is It?

Copilot Remote Control is a VS Code extension that streams your entire Copilot agent session to your phone in real time.

When the agent pauses and asks "Should I proceed?" — your phone buzzes. You glance at it, tap "yes," pocket it, and keep walking.

The agent never stops. You never have to be at your desk.

That's it. That's the whole product.

You → Give Copilot a big task → Walk away
Agent works → Hits a question → Your phone buzzes
You tap "yes" from your phone → Agent continues
You come back → Everything is done

Enter fullscreen mode Exit fullscreen mode


How It Works (The Technical Bit)

For the devs who care about architecture (I know you do):

1. Extension hooks into Copilot's JSONL output

The extension monitors Copilot agent mode's conversation in real-time using a lightweight 20ms JSONL polling mechanism. No modification to Copilot itself — it reads the existing output stream.

2. WebSocket server on localhost

A local WebSocket server spins up on your machine. Your code never touches any external server.

3. Cloudflare Tunnel for secure access

A Cloudflare tunnel creates a secure wss:// connection from your phone to your local machine. Peer-to-peer. No relay server storing your data.

4. React PWA on your phone

Scan a QR code from the VS Code sidebar → your phone opens a Progressive Web App that looks and feels like VS Code's dark theme. Full chat history, code snippets, even file sending.

5. Firebase Push Notifications

When the agent pauses, your phone gets a push notification via Google FCM. Even if the PWA tab is closed.

┌─────────────────┐         ┌──────────────┐         ┌─────────────────┐
│   VS Code +     │  wss:// │  Cloudflare  │  wss:// │   Your Phone    │
│   Copilot Agent ├────────►│   Tunnel     ├────────►│   React PWA     │
│   Extension     │         │  (no relay)  │         │   Push Notifs   │
└─────────────────┘         └──────────────┘         └─────────────────┘
     Your machine              Encrypted               Your pocket
     Your code stays here      Pass-through             Full control

Enter fullscreen mode Exit fullscreen mode

Your code never leaves your machine. Period.


The Response Has Been Incredible

I launched quietly a few weeks ago. No Product Hunt. No Twitter thread. Just published on the VS Code Marketplace and shared with a few developer communities.

Here's what happened:

📊 The Numbers

Metric Value
Marketplace installs Growing daily
Average session length 3+ hours
Most common response time Under 15 seconds
Refund requests Zero

💬 What Developers Are Saying

"This is one of those tools where you wonder why it didn't exist before. I gave Copilot a 200-file migration task, went for a walk, and responded to 4 prompts from my phone. Came back to a finished PR."

"I used to sit at my desk babysitting Copilot for hours. Now I just glance at my phone when it buzzes. It's changed how I think about AI coding."

"The fact that it's $5 one-time and not another $20/month subscription is insane. Respect."

"Set it up in 2 minutes. QR code scan, done. My phone is now my Copilot remote. This is the future."

What surprised me most:

People aren't just using it for quick "yes/no" responses.

They're writing full multi-line instructions from their phone while commuting. They're reviewing code diffs on the train. They're sending reference files from their phone to their VS Code workspace.

One developer told me they now start a big Copilot task before their morning commute and guide it from the bus. By the time they sit down at their desk, the work is done.

That's not how I designed it. But that's how people are using it. And I love it.


Why $5?

I know what you're thinking. "What's the catch?"

No catch.

  • $5 one-time. Not $5/month. Not $5/year. One payment. Done forever.
  • No cloud account. No sign-up. No tracking.
  • Offline license verification. RSA-2048 signed JWT. Works on air-gapped machines.
  • Unlimited devices. One license, every machine you own.

Why so cheap? Because I built this to solve my own problem. The fact that other developers find it useful is the bonus. I'd rather 10,000 developers use it at $5 than 100 developers use it at $20/month.

Also — Copilot already costs $10-19/month. Adding another subscription on top feels wrong. You're already paying for the AI. This just makes it actually usable when you're not at your desk.


The Setup (< 2 Minutes)

# Step 1: Install from VS Code Marketplace
ext install atulhritik.copilot-remote

# Step 2: Activate your license
Cmd+Shift+P → "Copilot Remote: Activate License"
# Paste your key → Done

# Step 3: Scan QR code
# Open the Copilot Remote sidebar → Scan the QR with your phone

# Step 4: Walk away
# Your phone is now connected. Every agent message streams live.

Enter fullscreen mode Exit fullscreen mode

That's it. No Docker. No config files. No environment variables. No "please install these 7 prerequisites."

Scan. Connect. Walk away.


What I Built It With

For the indie hackers and builders curious about the stack:

  • VS Code Extension: TypeScript, VS Code Extension API, JSONL file watcher
  • WebSocket Server: Node.js, runs locally on your machine
  • Tunnel: Cloudflare Argo Tunnel (free tier)
  • Phone PWA: React, WebSocket client with auto-reconnect, 200-message history buffer
  • Push Notifications: Firebase Cloud Messaging (FCM)
  • License System: RSA-2048 JWT, offline verification, no phone-home
  • Payment: Simple one-time checkout via Razorpay
  • Hosting: Vercel (PWA + license portal)

Built the entire thing end-to-end: architecture, extension, PWA, payment flow, deployment, and marketplace publishing. Solo.


What's Next

Based on feedback, here's what I'm working on:

  • [ ] Multi-agent support — Control multiple Copilot sessions from one phone
  • [ ] Voice replies — Whisper-powered speech-to-text for phone responses
  • [ ] Wearable notifications — Apple Watch / WearOS companion
  • [ ] Team mode — Share agent sessions with teammates
  • [ ] Session history — Browse past agent conversations from your phone

Try It

If you use GitHub Copilot agent mode and you've ever lost time waiting for a "Proceed?" prompt — this is for you.

🧩 Install from VS Code Marketplace

🔑 Get your $5 lifetime license

📱 Preview the phone PWA


Built by Atul Kr Srivastava — a developer who got tired of babysitting his AI agent.

If you found this useful, a ❤️ or share means the world. And if you have feedback or feature requests, hit me up — I read every message.