vercel-sandbox

Run headless Chrome browser automation inside Vercel Sandbox microVMs.

63|12|Updated Mar 9, 2026
One-click install
npx skills add https://github.com/iBz-04/gloamy --skill vercel-sandbox-ibz-04
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: vercel-sandbox
Source: https://github.com/iBz-04/gloamy/tree/main/.agents/skills/vercel-sandbox
Command: npx skills add https://github.com/iBz-04/gloamy --skill vercel-sandbox-ibz-04

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the problem of running browser automation (headless Chrome + agent-browser) inside a Vercel app without hitting local binary size limits or risking polluting the main server environment.

Core Features & Use Cases

  • Ephemeral microVM browser automation: Executes agent-browser commands inside Vercel Sandbox microVMs that spin up on demand and shut down after use.
  • Headless Chrome without heavy deployment coupling: Lets you perform tasks like scraping, page interaction, screenshots, and accessibility snapshotting without bundling large browser binaries into your main app.
  • Persistent browser sessions across commands: Enables multi-step workflows where the sandbox persists between sequential command invocations during the automation run.
  • Fast production startup via sandbox snapshots: Uses pre-built Sandbox snapshots (with system dependencies and agent-browser installed) to reduce cold-start time.

Quick Start

Ask your AI to run agent-browser in a Vercel Sandbox microVM to open a URL and return both the page title and a base64-encoded screenshot.

Frequently Asked Questions about vercel-sandbox

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run headless Chrome automation inside a Vercel app without hitting deployment size limits?

You can run headless Chrome automation inside a Vercel app by executing agent-browser commands within Vercel Sandbox microVMs, which avoids bundling large browser binaries into your main deployment and isolates the environment.

Can I take screenshots and capture accessibility snapshots on Vercel using Next.js or SvelteKit?

Yes, you can capture screenshots and accessibility snapshots by running headless Chrome inside Vercel Sandbox microVMs, supporting frameworks like Next.js, SvelteKit, Nuxt, Remix, and Astro for server-side tasks.

How do I execute multi-step browser form interactions across persistent sessions on Vercel?

You can execute multi-step form interactions by creating a Vercel Sandbox microVM that maintains a persistent browser session across sequential agent-browser command invocations during the automation run.

Do I need the @vercel/sandbox SDK and Vercel credentials to start an isolated browser session?

Yes, starting an isolated browser session requires the @vercel/sandbox SDK and Vercel credentials via OIDC or VERCEL_TOKEN, VERCEL_TEAM_ID, and VERCEL_PROJECT_ID to create, run, and stop microVMs.

What is the best way to reduce cold-start time for headless Chrome in Vercel Sandbox?

The best way to reduce cold-start time is to use an optional pre-built AGENT_BROWSER_SNAPSHOT_ID, which restores a Vercel Sandbox microVM with system dependencies and agent-browser already installed for sub-second startup.

Why use Vercel Sandbox microVMs instead of running browser automation directly in my server environment?

Using Vercel Sandbox microVMs prevents polluting your main server environment with browser dependencies and bypasses local binary size limits, spinning up on demand and shutting down safely after automation completes.