Turn any website into agent-ready context.
Point it at a URL, get back clean, structured markdown built for agents, not browsers. A free tier ships bundled with th; paid tiers unlock the volume real crawling needs.
$ th crawl https://example.com/pricing
# → clean markdown, ready for an agent's context
{ "markdown": "# Pricing…", "links": [ … ] }Structured for agents, not for eyeballs.
Clean, not raw HTML
Boilerplate, nav, and ad chrome stripped — just the content an agent needs, as markdown or structured JSON.
Renders the real page
Client-rendered sites resolve to their loaded content, not an empty shell. Paid tiers add higher-concurrency rendering.
Fast, resilient backend
Powered by search.smoo.ai — Brave/Exa/Parallel routing, Redis-cached, org-scoped by JWT.
Free where you start. Metered where you scale.
- Bundled with th — nothing to sign up for
- Public client key, edge rate-limited
- Clean markdown + structured JSON
- Authenticated with your org key (th api login)
- Metered overage, billed per org
- Higher concurrency + JS rendering
- Dedicated crawl pool + SLA
- Private routing, org-scoped isolation
- Volume pricing — talk to us
Preview— the crawl endpoint isn't generally available yet, and the limits and prices above are proposed, not final.
The bundled key can't touch your org.
The free-tier credential shipped in th is a public, scoped client — not a secret org key. It reaches the crawl route and nothing else.
Crawl route only
The bundled client belongs to a dedicated, minimally-scoped org — never CRM, config, or knowledge. Extract it and you get free crawl, full stop.
Abuse-proofed at the edge
Rate limits key on IP + install id, not on the key being secret — so a public key stays a public key.
Paid = your identity
Pro and Enterprise authenticate with your own org (th api login), metered and billed per org. The shared key never unlocks paid.