Firecrawl scrape formats: the seven Sume supports

Sume's scrape formats array accepts seven values: markdown (the default), links, html, json, screenshot, rawHtml and images. Firecrawl's scrape reference lists more, including Summary, Branding, Product, Menu, Audio, Video, Question and Highlights. Those have no Sume equivalent.

Vendor list from Firecrawl's scrape reference, Sume's from its OpenAPI enum and MCP tool description, read 2026-09-30.

Which Firecrawl formats exist on Sume?

The names that match are Markdown, HTML, Raw HTML, Links, Images, Screenshot and JSON. Sume spells them markdown, html, rawHtml, links, images, screenshot and json.

What if I relied on summary or question?

Use the json format, which is Sume's structured-answer path. A json_prompt of up to 2,000 characters can ask for a summary or one answer; see the question-format post. For brand colors or product fields there is nothing to switch on, so describe the fields in json_schema and read them from the page.

What else shapes the output?

only_main_content defaults to true, so navigation and footers are trimmed. timeout_ms goes up to 30000, wait_for_ms from 0 to 10000, and actions takes up to 8 steps. Output is bounded and a truncated marker says when it was cut; page content is untrusted.

How many formats can one request ask for?

formats takes 1 to 7 unique values, so one call can return markdown and links together. Ask only for what you will read: rawHtml and html are large.

const res = await fetch("https://api.sume.com/v1/firecrawl/scrape", {
  method: "POST",
  headers: {
    Authorization: "Bearer " + process.env.SUME_API_KEY,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ url: "https://example.com", formats: ["markdown", "links"] }),
});
console.log(res.status, await res.json());