Vercel Sandbox

Vercel Sandbox is an ephemeral compute primitive designed to safely run untrusted or user-generated code on Vercel. It supports dynamic, real-time workloads for AI agents, code generation, and developer experimentation.

Key Use Cases:

• Execute Untrusted Code: Safely run code from unverified sources, such as AI agent outputs or user uploads, without risking production systems.
• Build Dynamic Experiences: Create interactive tools that generate or modify code on the fly, like AI-powered UI builders or developer playgrounds.
• Isolated Backend Testing: Preview and test user-submitted or agent-generated code in a self-contained environment with access to logs and live previews.
• Run Development Servers: Test applications in an isolated environment.

Features:

• Supported Runtimes: Includes node22 and python3.13 images based on Amazon Linux 2023.
• Package Management: Supports npm, pnpm for Node.js and pip, uv for Python. Users can also install additional system packages using dnf.
• Authentication: Integrates with Vercel OIDC tokens for seamless authentication or supports Vercel access tokens for other environments.
• Observability: Track sandbox usage, inspect command history, and view logs through the project's Observability tab in the Vercel dashboard.