/
Documentation

Overview

What is snapwright?

snapwright is a persistent browser server you control from the command line. Start it once, then send it commands — navigate, click, fill forms, read page content, take screenshots — without reopening a browser each time.

It’s built for workflows where you need to automate or observe a web UI repeatedly: manual testing, documentation, AI-driven automation, or capturing screenshots for reports.

How it works

snapwright start          ← launches a Chromium browser in the background
snapwright navigate ...   ← sends commands to that browser over HTTP
snapwright screenshot ...
snapwright stop

The browser stays open between commands, so cookies, session state, and page context are preserved across the entire workflow.

Key features

  • Persistent session — the browser doesn’t restart between commands; state is preserved
  • Auth persistence — log in once, save the session, restore it on future runs
  • Full-page screenshots — captures the entire page or clips to a specific element
  • Automatic PNG compression — screenshots are compressed ~60–70% in size
  • AI agent integration — works out of the box with Claude Code via built-in skills

Quick start

# 1. Install
npm install -g @meldocio/snapwright
snapwright install-browsers

# 2. Start the browser
snapwright start

# 3. Use it
snapwright navigate https://example.com
snapwright screenshot output/home.png
snapwright stop

What’s next