π Tutorial: From Zero to Live App in 5 Minutes (The AI Agent Stack)
I heard of GitHub CodeSpaces where I donβt need a local machine to develop code. Then I also heard I could use Claude Code as a coding agent. So, I connected them both and had Claude create a very simple webapp for proof of concept of this dev environment, and it worked. It is just so very nice so I am sharing my experience of it in a tutorial format. Below is what I had Gemini 3 write. I tested the process and it works. Enjoy!
For context, here is the deployed page in 5 minutes or so (mostly for dev configuration.)
https://pomodoro-timer-demo-five.vercel.app/
This guide outlines a modern "Cloud-Native" workflow using GitHub Codespaces and Claude Code. It eliminates local environment setup, allowing you to go from an empty repository to a deployed app on your phone in minutes.
The Stack
Environment: GitHub Codespaces (Cloud Linux VM)
Developer: Claude Code (AI Agent in the terminal)
Deployment: Vercel (Continuous Deployment)
Step 1: The "Instant" Computer (GitHub Codespaces)
Instead of setting up your messy local machine, we use a clean cloud environment.
Go to GitHub and create a new empty repository (e.g.,
ai-speed-run).Click the green Code button -> Codespaces tab -> Create codespace on main.
Result: In ~30 seconds, you have a full VS Code editor running in your browser.
Step 2: Hire the Agent (Claude Code)
We invite the AI into the terminal to write the code for us.
In the Codespaces terminal, install the agent:
Bash
npm install -g @anthropic-ai/claude-codeLog in (follow the link it provides):
Bash
claude loginResult: You now have an AI developer ready to accept commands.
Step 3: The Build
Don't write code. Describe the outcome.
Run the agent:
Bash
claude "Create a Cyberpunk Pomodoro Timer in a single index.html file with internal CSS/JS. Make it look like a matrix terminal."Result: Claude creates the files and writes the logic instantly.
Step 4: The Preview
Verify the app works inside the cloud.
Run a simple Python server (pre-installed in Codespaces):
Bash
python3 -m http.server 8080Click the "Open in Browser" popup.
Result: You see your live app running on a private URL.
Step 5: The "Forever" Deploy (Vercel)
Make it real.
Tell Claude to save the work:
Bash
claude "Git add, commit, and push to main"Go to Vercel.com -> Add New Project -> Select your repo -> Deploy.
Result: You get a live URL (e.g.,
timer.vercel.app) that updates automatically every time you ask Claude to push changes.
