Loading...
Loading...
Work with Vercel Sandbox — ephemeral Linux microVMs for running untrusted code, AI agent output, and developer experimentation on Vercel. Use this skill when the user mentions "Vercel Sandbox", "@vercel/sandbox", sandbox microVMs, running code in isolated environments on Vercel, or wants to create/manage/snapshot sandboxes via the TypeScript/Python SDK or Vercel CLI. Also trigger when the user asks about sandbox pricing, resource limits, authentication (OIDC tokens, access tokens), system specifications, CLI commands (`vercel sandbox`), or wants to update the local documentation cache for this skill.
npx skill4agent add cesarecaoduro/agent-skills vercel-sandboxpnpm i @vercel/sandbox # TypeScript
pip install vercel-sandbox # Pythonvercel link # Link to a Vercel project
vercel env pull # Pull OIDC token to .env.local (expires 12h locally)import { Sandbox } from '@vercel/sandbox';
const sandbox = await Sandbox.create();
const result = await sandbox.runCommand('echo', ['Hello from Vercel Sandbox!']);
console.log(await result.stdout());
await sandbox.stop();| Class | Purpose |
|---|---|
| Create, manage, list, snapshot microVMs |
| Interact with running/detached commands |
| Access exit code, stdout, stderr after completion |
| Save/restore sandbox state for fast restarts |
node24node22python3.13/vercel/sandboxvercel-sandboxsandbox.extendTimeout()sandbox.snapshot()web_fetchweb_fetchweb_fetch/docs/vercel-sandbox/references/> Last fetched:python3 <skill-path>/scripts/update_docs.py <skill-path>/references