Length you control
Set max_sentences and get a summary sized for your prompt, from a one-liner to a full paragraph.
POST /v1/summarize
Page summaries your agents can read.
Collapse any page into a tight summary with a single call.
Feed your LLM the gist, not the whole page. Send a URL, get back a clean summary your agents can act on without burning context on nav, ads, and boilerplate. Built in Rust, scrapes and cleans the page before the model ever sees it, and you control the length down to the sentence.
What you get
Everything in one call.
Clean source content
The page is scraped and stripped of nav, ads, and boilerplate before the model reads it, so the summary stays on topic.
Token savings built in
Hand your LLM a summary instead of raw HTML and cut roughly 90% of the tokens you would otherwise spend on the page.
Handles the hard pages
JavaScript rendering and bot protection are dealt with automatically, so summaries come back even from pages a plain fetch can't load.
How it works
From URL to output in four steps.
01
Send a URL
POST the page URL and an optional max_sentences to set how long the summary should be.
02
Page gets fetched
Webclaw scrapes the page in Rust, rendering JavaScript and clearing bot protection when needed.
03
Content gets cleaned
Nav, ads, and boilerplate are stripped so only the real content is passed to the model.
04
Summary returned
An LLM condenses the cleaned text and you get back a single structured summary field.
API
One request, structured back.
POST /v1/summarize
Send a POST to /v1/summarize with the page URL and an optional max_sentences. Webclaw scrapes and cleans the page, then an LLM returns a structured summary field. No scraping or prompt setup on your side.
Ship an agent that actually sees the web.
One credit pool, every endpoint. Cancel anytime, or self-host the open-source core for free.