user@gryczka.dev: ~/garden/browser-rendering
$ cat ~/garden/browser-rendering/README.md
Browser Run API Explorer screenshot

Browser Run API Explorer

October 30, 2025 | Patrick Gryczka

## Overview

Try Browser Run on a public URL and inspect the response. The snapshot explorer combines two to eight outputs from one page load, including CSS matches and AI-extracted JSON. A saved capture lets you inspect and download all eight outputs without using the public demo allowance. The crawl explorer shows each URL’s outcome and the robots.txt status for each reported origin. A separate comparison runs Markdown extraction through both the REST API and a Workers browser binding. The Worker keeps the API token server-side and uses per-IP rate limits, crawl ceilings, and a shared Durable Object budget to bound public usage.

## Highlights

  • -> Covers all ten Browser Run Quick Actions available over REST
  • -> Eight snapshot formats with per-output previews and downloads
  • -> Async crawl results with per-URL outcomes and per-origin robots.txt diagnostics
  • -> Side-by-side Markdown extraction over REST and the Workers binding
  • -> Public URL validation, per-IP rate limits, crawl ceilings, and a shared daily budget
  • -> Browser time and output size alongside the capture

## Architecture

Browser Run architecture diagram
Browser (React UI + TanStack Router)
    │   Server Function Call (RPC over HTTP)
    ▼
Cloudflare Worker (TanStack Start)
    │   REST API Call (Bearer token auth)
    ▼
Browser Run Quick Actions API
    │   Spawns headless Chrome session
    ▼
Headless Chrome (Cloudflare Network)
    │   Navigates & renders target websites
    ▼
Result (screenshot, PDF, HTML, JSON, Markdown...)

## Tech Stack

WorkersBrowser RunDurable ObjectsRate LimitingTanStack StartTypeScript
$ _