A faithful summary of any URL, in Markdown

The request

One POST, one bearer token

curl -X POST https://urlpipe.dev/summarize \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com/blog/launch", "sync": true}'

import requests

res = requests.post(
    "https://urlpipe.dev/summarize",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={"url": "https://example.com/blog/launch", "sync": True},
)

const res = await fetch("https://urlpipe.dev/summarize", {
  method: "POST",
  headers: {
    "Authorization": "Bearer YOUR_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ url: "https://example.com/blog/launch", sync: true }),
})

$ch = curl_init("https://urlpipe.dev/summarize");
curl_setopt_array($ch, [
    CURLOPT_POST => true,
    CURLOPT_RETURNTRANSFER => true,
    CURLOPT_HTTPHEADER => [
        "Authorization: Bearer YOUR_API_KEY",
        "Content-Type: application/json",
    ],
    CURLOPT_POSTFIELDS => json_encode(["url" => "https://example.com/blog/launch", "sync" => true]),
]);

$response = curl_exec($ch);

The post announces version 3 of the product, which moves rendering into
the browser and drops the separate export step.

It walks through the three changes users will notice: faster previews,
**offline editing**, and a new sharing dialog. Existing projects open
unchanged; the old export command keeps working until the next release.

It ends with pricing, which stays the same, and a link to the migration notes.

The summary

What the summary is held to

• The whole page, not its opening. Every major section gets at least a sentence, in the order the page takes them, through to what it ends with.
• Only what the page says. No facts, context or opinions added, and no detail the page does not state.
• The page's own content. Reader comments, reviews, ads and the other pages a site suggests alongside are left out.
• Three to five paragraphs of Markdown, with emphasis and lists where the content calls for them, and no heading, label or preamble to strip.
• In the page's language. A German article is summarized in German.

Use cases

What people build with it

• Digests. A newsletter, a Slack channel or a daily briefing built from a list of links.
• Link previews with substance. A paragraph about what a shared page actually says, beside the title and image from the metadata API.
• Research triage. Decide which sources deserve a full read before anyone opens them.
• Short context for a model. A summary when the whole page would crowd the context window; the Markdown API when the model needs every word.

import requests

links = ["https://example.com/a", "https://example.com/b"]
digest = []
for url in links:
    res = requests.post(
        "https://urlpipe.dev/summarize",
        headers={"Authorization": "Bearer YOUR_API_KEY"},
        json={"url": url, "sync": True},
    )
    digest.append(f"## {url}\n\n{res.text}")

print("\n\n".join(digest))

Limits

What it does not do

• No length or style setting. Three to five paragraphs, in the page's language; there is no word count, bullet-only mode or prompt of your own.
• No pages behind a login or a hard paywall: it summarizes what the page shows a visitor.
• One page per request. It does not follow links or summarize a whole site.
• Up to 10 MB of HTML, and public pages only.