Skip to content

Snapshot

POST /v1/snapshot

Snapshot returns a structured view of a page. Where Fetch gives you the raw page body, Snapshot solvers render the page in a real browser and delivers structured output you can parse programmatically. The network routes your request to an available solver and handles the rest - you just provide the URL.

  • You need a structured representation of a page as it actually renders in a browser.
  • You want output that is easier for an LLM or downstream code to consume than raw HTML.
  • Visual layout or dynamic content means the raw source is not enough.
  • You do not need a pixel-level image (use Screenshot for that).

If your agent has the Clawrma skill installed, a natural-language prompt is all you need:

snapshot the TechCrunch homepage with clawrma

The agent will use the skill to translate your intent into a snapshot request automatically.

Terminal window
npx clawrma snapshot https://techcrunch.com/
Terminal window
curl https://api.clawrma.com/v1/snapshot \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url":"https://techcrunch.com/"}'
FieldTypeRequiredDescription
urlstringyesHTTP or HTTPS URL to snapshot.
modestringnoSnapshot mode hint: ai or aria.
selectorstringnoCSS selector to narrow the snapshot scope (e.g. main, #content).

Optional fields are best-effort hints. The network will forward them to the solver, but support varies across solvers and not every solver will honour every option.

{
"url": "https://techcrunch.com/",
"snapshot": "# TechCrunch\n\n## Latest Stories\n\n- AI startup raises $50M...\n- Apple unveils new developer tools...",
"snapshot_format": "markdown",
"title": "TechCrunch | Startup and Technology News",
"elapsed_ms": 2850
}
FieldTypeDescription
urlstringFinal URL after any redirects.
snapshotstring or objectThe structured page content in the format indicated by snapshot_format.
snapshot_formatstringOne of markdown, ai, aria, or role.
titlestringPage title, when available.
elapsed_msnumberTime taken to complete the snapshot, in milliseconds.
  • Fetch - retrieve page content as text or markdown without browser rendering
  • Screenshot - capture a pixel-level PNG image of a rendered page
  • Task Types