Output

Overview & Purpose

output is the cross-cutting control that shapes what comes back from any scrape-capable endpoint — raw markup, a cleaned/converted version, a link list, an image, or extracted structured data. It works the same way everywhere it appears, so understanding it once means you understand it on /scrape, /batch/scrape, /crawl, and /serp. Use this page to decide which format(s) to request and how JSON extraction works — for the exact request/response field types, see each endpoint’s API reference page. Prerequisites: none beyond a valid API key — output is a plain request field, not a separate authenticated flow.

Supported formats

`output` value	Response key	What it is
`html`	`html_body`	Raw HTML of the page, exactly as fetched.
`clean_html`	`clean_html`	HTML with boilerplate stripped — see `only_main_content` below.
`markdown`	`markdown`	The page converted to Markdown.
`links`	`links`	Every link found on the page, as a flat list.
`screenshot`	`screenshot`	An image of the visible viewport.
`screenshot_full_page`	`screenshot_full_page`	An image of the entire scrollable page.
`json`	`json`	Extracted structured data — see below.

Request several formats at once — e.g. output: ["markdown", "links"] — and each comes back under its own key in data, computed from the same single fetch.

JSON extraction: `prompt` and `schema`

When output includes json, pline reads the rendered page and extracts structured data guided by prompt (a plain-language instruction) and/or schema (the exact shape you want back). Use one or both:

prompt alone — fast, loosely-structured extraction when field names don’t need to be exact.
schema alone — a fixed, typed shape your code can rely on without guessing keys.
Both together — schema pins the shape, prompt disambiguates content the schema can’t (e.g. which of several prices is the current one). This is usually the most reliable combination.

Best practices

only_main_content (default true) strips navigation/footers/ads from clean_html and markdown only — it has no effect on html, links, screenshot, or json.
screenshot and screenshot_full_page are mutually exclusive — requesting both is rejected.
Not every endpoint supports every format — Crawl’s per-page output has no json or screenshot_full_page; /serp enrichment only supports html/clean_html/links. Check the endpoint’s own reference page before assuming full parity with /scrape.
Combine formats instead of making two requests — asking for markdown and json together costs one fetch, not two.

Practical Implementation Example

Scenario: get a readable copy of an article for your own archive and a structured summary in a single request.

curl -X POST "https://api.scrapeswitch.com/scrape" \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://news.example.com/article/123",
    "output": ["markdown", "json"],
    "prompt": "Extract the headline, author, and publish date.",
    "schema": { "headline": "string", "author": "string", "published_at": "string" }
  }'

import os
import httpx

response = httpx.post(
    "https://api.scrapeswitch.com/scrape",
    headers={"Authorization": f"Bearer {os.environ['API_KEY']}"},
    json={
        "url": "https://news.example.com/article/123",
        "output": ["markdown", "json"],
        "prompt": "Extract the headline, author, and publish date.",
        "schema": {"headline": "string", "author": "string", "published_at": "string"},
    },
    timeout=60.0,
)
data = response.json()["data"]
archive_copy = data["markdown"]
summary = data["json"]

const response = await fetch("https://api.scrapeswitch.com/scrape", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    url: "https://news.example.com/article/123",
    output: ["markdown", "json"],
    prompt: "Extract the headline, author, and publish date.",
    schema: { headline: "string", author: "string", published_at: "string" },
  }),
});

const { data } = await response.json();
const archiveCopy = data.markdown;
const summary = data.json;

Error codes

Code	When you receive it
`422`	`output` includes `json` without a `prompt` or `schema`, or requests `screenshot` and `screenshot_full_page` together, or requests a format the target endpoint doesn’t support.

Availability by endpoint

Endpoint	Supported `output` values
`POST /scrape`	All seven: `html`, `clean_html`, `links`, `markdown`, `screenshot`, `screenshot_full_page`, `json`
`POST /batch/scrape`	All seven, applied to every URL in the batch
`POST /crawl`	`markdown`, `html`, `clean_html`, `links`, `screenshot` — no `json` or `screenshot_full_page`
`GET /serp`	`html`, `clean_html`, `links` only, as optional result enrichment

Get started

Endpoints

Teams

Billing

Recipes

Monitoring

Overview & Purpose

Supported formats

JSON extraction: `prompt` and `schema`

Best practices

Practical Implementation Example

Error codes

Availability by endpoint

Reference Links

​Overview & Purpose

​Supported formats

​JSON extraction: prompt and schema

​Best practices

​Practical Implementation Example

​Error codes

​Availability by endpoint

​Reference Links

Overview & Purpose

Supported formats

JSON extraction: `prompt` and `schema`

Best practices

Practical Implementation Example

Error codes

Availability by endpoint

Reference Links