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

D
DataBreaches.Net
IT之家
IT之家
The Cloudflare Blog
Apple Machine Learning Research
Apple Machine Learning Research
WordPress大学
WordPress大学
N
Netflix TechBlog - Medium
阮一峰的网络日志
阮一峰的网络日志
P
Proofpoint News Feed
L
LangChain Blog
博客园 - Franky
美团技术团队
J
Java Code Geeks
Microsoft Security Blog
Microsoft Security Blog
博客园 - 叶小钗
小众软件
小众软件
Y
Y Combinator Blog
B
Blog RSS Feed
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
D
Docker
Hugging Face - Blog
Hugging Face - Blog
Jina AI
Jina AI
罗磊的独立博客
大猫的无限游戏
大猫的无限游戏
Vercel News
Vercel News

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 USB drive that runs an offline AI coding agent ...
Muhammad Usm · 2026-05-13 · via DEV Community
Cover image for I built a USB drive that runs an offline AI coding agent on any laptop

Muhammad Usman

I wanted something like Claude Code — but completely offline, on any laptop,
with nothing installed on the host machine.

So I built code-stick.

What it is

A CLI that turns any USB drive into a portable, self-contained AI coding agent.
You get file editing, multi-step tasks, and tool use — running entirely on the
stick against a local model. No internet. No installs. Nothing left on the host
when you unplug.

How it works

Run this once on your own machine:

npx code-stick install

Enter fullscreen mode Exit fullscreen mode

Pick a model — Qwen2.5-Coder, DeepSeek-Coder, CodeGemma, or Phi-3.

It downloads everything onto the USB:

  • opencode + Ollama pre-built binaries for all five targets
  • Model weights in a USB-local Ollama store
  • Three launchers at the root: start-windows.bat, start-mac.command, start-linux.sh

From that point on, plug into any Windows, macOS, or Linux laptop, double-click
the matching launcher, and you get a full coding agent running at 127.0.0.1.

Quit → Ollama process killed by PID → host stays completely clean.

Why I built it

Three situations cloud agents can't serve:

Airgapped environments — banks, hospitals, defense, lab VMs where you can't
install software or call external APIs.

Shared or borrowed machines — school computers, client laptops, internships
where the host must stay untouched.

Privacy-sensitive code — NDA'd or client code that legally can't go to a
hosted model.

Current state

v0.1.0 — early but working:

  • Tested end-to-end on all five targets
  • Docker-based smoke test in CI + Vitest unit suite
  • Crash reports are local-only, no auto-upload, no telemetry
  • MIT licensed, Node 20+

Try it

GitHub: github.com/MuhammadUsmanGM/code-stick

Would love feedback — especially on model selection, USB performance, and any
edge cases on your OS.