testingurl

testingurl.dev is a free sandbox of pages with known, predictable structures for practicing web scraping, browser automation, and HTTP client development.

Every page returns the same content on every request unless its own description says otherwise, so it's safe to write automated tests and scrapers against. robots.txt allows every path (crawlers and AI agents included), llms.txt gives agents a structured index of every page, and every JSON endpoint is documented in a validated OpenAPI 3.0 spec.

Site structure

• /scraping — an ecommerce catalog (category → product pages), product data in JSON-LD/Microdata/Open Graph/dataLayer, pagination patterns (with rel="next"/"prev" signals), JS-rendered and iframe-nested content, tables, forms (including a multi-step wizard and hidden/honeypot fields), a cookie-session login wall, link graphs, elements with randomized ids and delayed state changes, paired accessible/inaccessible pages for testing a11y scanners, the same content rendered at several markup difficulty levels, URL normalization edge cases (trailing slash, case sensitivity, query param order, the <base> tag), canonical tag and meta robots/X-Robots-Tag scenarios, hreflang variants, and non-HTTP redirect mechanisms (meta refresh, JS redirect, an intentional infinite loop)
• /http — status codes, redirects, header/cookie inspection, request echoing, a User-Agent/header-spoofing check, basic/bearer auth, gzip, rate limiting, conditional GET, streamed responses, and byte-range requests
• /mock-data — deterministic fake REST APIs for Users, Posts & Comments, Albums & Photos, and Todos
• /generator — on-demand, non-deterministic generators for bulk fake users, test payment card numbers, dummy files, and placeholder images
• /contact — contact details exposed as plain text, Microdata, and JSON-LD, for testing contact-data extraction
• /ai — fixtures for people building AI agents and RAG pipelines. A prompt-injection test page has seven labeled hidden-text vectors plus a benign canary phrase. A retrieval corpus (HTML + Markdown + PDF + bulk JSON, plus a single dataset.json bundle) comes with a documented query → expected-result answer key. There are also dedicated PDF documents built to stress-test extraction: a real table, a two-column layout, and repeated running headers/footers

Stack

• Hono (TypeScript)
• Server-rendered HTML via hono/jsx

Local development

npm install
npm run dev

Starts a local dev server at http://localhost:8787.

To exercise /http/rate-limit, create a KV namespace and set its id in wrangler.toml:

wrangler kv namespace create TESTINGURL_KV

Without a bound namespace, that one endpoint returns 501. Every other page works without it.

Deploying

npm run deploy

Then attach your custom domain from the Cloudflare dashboard (Workers & Pages → your worker → Settings → Domains & Routes), or:

wrangler deploy --routes yourdomain.com/*

License

MIT