Paste any confusing website and get a clear, jargon-free explanation of what it actually is and does. Every URL you explain becomes a permanent, server-rendered, cached page at siteexplainer.com/<url> (e.g. siteexplainer.com/vercel.com).

1. You paste a URL (or go straight to siteexplainer.com/<url>).
2. We fetch the page's own HTML and distill it down to the meaningful text.
3. We send that to OpenRouter — by default the free models (openrouter/free + a fallback list of free instruct models) — to write a short, plain-English explanation grounded only in the page's real content.
4. The result is cached in Redis forever and rendered server-side, so the page loads instantly and is fully indexable.

Because generation runs on free models, there's no per-request cost — OpenRouter's own free-tier limits are the rate limiter, so the app has no rate-limiting code.

• Next.js 16 (App Router, React 19, server components + streaming)
• Tailwind CSS v4
• OpenRouter free models for generation (clean instruct models with automatic fallback when one is rate-limited; pin one via OPENROUTER_MODEL)
• Upstash Redis — caches explanations and the "recently explained" list
• Deployed on Vercel

Copy .env.example to .env and fill in the keys:

cp .env.example .env

• OPENROUTER_API_KEY — from https://openrouter.ai/keys (required)
• UPSTASH_REDIS_REST_URL / UPSTASH_REDIS_REST_TOKEN — optional cache; create a free DB at https://console.upstash.com
• OPENROUTER_MODEL — optional, pin a single model instead of the free fallback list
• NEXT_PUBLIC_SITE_URL — your public origin (used for canonical/OG + sitemap)

Then:

npm install
npm run dev

The app runs at http://localhost:3000. Without Redis credentials it still works — it just won't cache (every visit regenerates).

The app degrades gracefully: if Redis is unavailable, explanations are still generated, just not cached. SSRF-hardened fetcher (DNS + private-IP checks on every redirect hop). SSR + per-page metadata, sitemap, and robots make every /<url> page indexable.