Written in Rust — public beta

Web retrieval without the cruft.

Rust API that cleans web content for AI agents — no headless browser, no bloat.

50 requests/month free when we launch. No spam.

Before / After

What Cruftless removes vs. what you get back.

a typical pagecruft

POST /v1/cleanExample output

{
  "title": "Web scraping - Wikipedia",
  "wordCount": 2847,
  "content": [



]}

Example output from a real fixture — not a live demo yet.

Why Cruftless

Different architecture. Different results.

No headless browser fleet

Pure HTTP fetch in Rust. No Chrome instances to run, patch, or pay for.

Transparent pricing

Every tier's cost per 1,000 requests is public. No sales call required.

Fails loudly, not silently

Specific error codes for every failure mode, never an empty response pretending to be success.

Pay for what you get

1 credit = 1 cleaned document returned, never billed per API call.

Simple, transparent pricing

Every plan includes the Cruft Cleaner. One credit = one cleaned document.

Starter$0

50 req/mo included

For integration testing

Included: 50/mo

Builder$29/mo

6,000 req included, then $4/1K

For side projects in production

Included: 6.0K/mo
Overage: $4/1K

Max$99/mo

25,000 req included, then $3.50/1K

Production agents, webhooks, email support

Included: 25.0K/mo
Overage: $3.5/1K

Pay-as-you-go$4.50/1K

Prepaid credits, valid 12 months, no commitment

For spiky or evaluation workloads

Included: 0/mo
Overage: $4.5/1K

All figures are estimates based on public pricing and typical usage. Actual pricing and savings may change at launch.

Questions

Honest answers. No sales fluff.

Does Cruftless work on any website?

Not yet on every site. Pages that render content with JavaScript (like Notion or X) won't work — that's a direct result of not using a headless browser, not a bug. Sites with aggressive bot-detection are hit-or-miss today, and that's an area we're actively improving. Works best on documentation, articles, blogs, and most news sites.

Why no headless browser?

Running a browser per page is where most of the cost and RAM usage comes from in similar tools. Skipping it keeps Cruftless fast and cheap — the tradeoff is not being able to render JS-only content.

What happens if a page can't be cleaned?

You get a specific error code back (no_content_extracted, fetch_failed) instead of a broken or silently empty response. Extraction quality across more site types is ongoing work.

How does billing work?

One credit equals one cleaned document returned, never one credit per API call. Search and get 8 results back — that's 8 credits, not 1.

When can I start using the API?

We're in public beta. Join the waitlist for early access — pricing and limits are already public above.

Do you publish performance benchmarks?

Not yet. We'll publish measured latency and token-reduction numbers once we've run them against a real benchmark set, not before.

Join the waitlist

Tell us who you are. We'll email you when it activates at launch.