Vercel Sandbox
Learn how to safely execute untrusted code using Vercel Sandbox. Build a CLI code review agent that clones repositories, runs tests, and uses AI to analyze code for security and quality issues, all in an isolated environment.
AI is writing more of your code. Claude Code oneshotting, ChatGPT copy and paste, forgetting what an IDE looks like. But running code you didn't write is risky. What if there are bugs? What if there are security vulnerabilities?
Before you YOLO everything into production, Vercel Sandbox lets you execute untrusted code in isolated microVMs that spin up in milliseconds. In this course, you'll build a CLI agent that doesn't just read code, it actually runs it, safely.
What you'll build
A CLI tool that accepts a GitHub repo URL, clones the repository into an isolated Sandbox environment, runs its test suite, and uses AI to review the code for security and quality issues. By the end, you'll have a tool that combines static analysis with dynamic test execution into a unified review.