POST /v1/summarize

Web page summarization API for digests and briefings

Turn any URL into a short, structured summary in one call.

Summarize any public web page into a clean digest with one API call. Feed automated briefings, change alerts, and newsletters without reading full articles or wiring up your own LLM.

Summarize a page

View API docs

How it works

Build it step by step.

The real flow, one step at a time. Switch between TypeScript, Python, and cURL on any snippet.

Send a URL

Call /v1/summarize with the page URL and an optional focus prompt describing what matters to you.

// Pass the page URL plus an optional focus prompt and lengthconst result = await wc.summarize({  url: "https://example.com/blog/quarterly-update",  focus: "product launches and pricing changes",  length: "short",});

2
webclaw fetches and cleans
The page is fetched with bot protection handled, then stripped of navigation, ads, and boilerplate before summarization.
3
Get a focused summary
The endpoint returns a concise summary sized to your chosen length, ready to read or store.
```
// The summary comes back sized to your chosen lengthconsole.log(result.summary);
```
4
Route it anywhere
Push the summary into a daily digest, a Slack briefing, an email newsletter, or a change alert.
```
// Drop the summary straight into a daily digestawait digest.add({  source: result.url,  summary: result.summary,});
```

Summarize a page

Why webclaw

Built for content summarization.

One call from URL to summary: no separate scrape, clean, and LLM steps

Focus prompt steers what the summary emphasizes (pricing, product, risks)

Reads bot-protected pages that other summarizers fail to fetch

Cached replay returns the same summary instantly for repeated URLs

Per-page pricing stays predictable across thousands of summaries

What you get

Everything this use case needs.

URL-to-summary in a single call
Focus prompt to steer the summary
Adjustable summary length
Bot-protected pages handled automatically
Cached replay for repeated URLs

Where it fits

Built for the messy parts.

Building a summary feature means scraping the page, stripping boilerplate, fitting the result inside a context window, calling an LLM, and parsing the output. Each step has its own failure modes, and modern bot protection breaks most scrapers before you ever reach the model.

webclaw /v1/summarize takes a URL, fetches and cleans the page in one pass, then returns a focused summary. You can steer it with a focus prompt and control length, so the same endpoint produces a one-line alert or a paragraph briefing. No separate scrape step, no token budgeting, no model wiring.

Common questions

Frequently asked questions

How is /v1/summarize different from scraping then calling an LLM myself?

It collapses the whole chain into one request. webclaw fetches the page, strips navigation and boilerplate, fits the content to the model, and returns a summary. You skip the scrape step, the token budgeting, and the prompt plumbing.

Can I control what the summary focuses on?

Yes. Pass a focus prompt to emphasize a topic, such as pricing changes or product announcements, and set the length to get a one-line alert or a longer briefing from the same endpoint.

What pages can I summarize?

Any public web page: news articles, blog posts, release notes, documentation, and changelogs. Pages behind common bot protection are fetched automatically, so summaries do not fail on protected sites.

For AI agents

Or hand it to your agent.

Add the webclaw MCP server to Claude, Cursor, or any MCP client, then paste this prompt. The agent calls the webclaw tools and hands the result back to your model — no code to write.

PROMPT FOR YOUR AGENT

Using the webclaw tools, summarize [the article or page URL] into a concise digest I can drop straight into my briefing. Call the summarize tool on that URL, and emphasize [what you want it to focus on, e.g. pricing changes and product announcements] while keeping it [short, e.g. a few sentences]. Return the result as a clean summary with the source URL above it and 3-5 bullet-point takeaways below it, so I can skim the key points without reading the full page. If I give you several URLs at once, summarize each one and lay them out as a table with columns for Source, One-line takeaway, and Summary.

Set up the MCP server