Web Reader API

Endpoint

POST https://serpforai.dev/api/v1/url
GET  https://serpforai.dev/api/v1/url

Authentication is the same bearer header used by the SERP API:

Authorization: Bearer <userKey>

This single endpoint also powers File Extraction and Screenshot - which response you get back depends only on which optional fields you set. Plain s gives you Reader; adding file: 1 switches to File Extraction; adding image: 1 or image: 2 switches to Screenshot.

Request parameters

Field	Type	Required	Default	Description
s	string	yes	-	The page URL to extract. When file: 1, this is the document URL instead
mode	number	no	0	Render mode: 0 = standard HTTP fetch, 1 = headless browser
w	number	no	5000	Wait after page load, in milliseconds. Only used when mode: 1
d	number	no	30000	Request timeout, in milliseconds
html	number	no	0	1 = include the raw page HTML alongside the Markdown, 0 = omit
proxy	number	no	0	Proxy tier: 0 none, 1 Shared Pool (+2 credits), 2 Datacenter (+5 credits), 3 Residential (+10 credits)
captcha	number	no	0	1 = attempt to solve simple CAPTCHAs automatically. No extra credits
file	number	no	0	File Extraction: 1 = parse the document at s (PDF/DOCX/XLSX/…). No extra credits
image	number	no	0	Screenshot: 1 = first screen only, 2 = full page. No extra credits

Response

{
  "code": 0,
  "msg": "",
  "data": {
    "id": "...",
    "markdown": "...",
    "title": "...",
    "description": "..."
  }
}

Field	Present when	Description
id	always	Opaque identifier for this extraction
markdown	always	The page content, converted to Markdown
title	always	Page title
description	always	Page description/summary
html	html: 1	Raw page HTML
fileUrl	file: 1	The original document URL, echoed back
fileMarkdown	file: 1	Parsed document content as Markdown
imageUrl	image: 1|2	Hosted PNG URL of the screenshot

When to enable the browser

Set mode: 1 only for pages that render their main content with client-side JavaScript. Headless rendering costs more time, so leaving mode: 0 (the default) keeps latency low for ordinary article pages.

curl -X POST https://serpforai.dev/api/v1/url \
  -H "Authorization: Bearer $SERPFORAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"s":"https://example.com/spa-article","mode":1,"w":2000,"d":15000}'

Proxies and CAPTCHAs

Most sites need nothing extra. Reach for proxy only when a target consistently blocks the platform’s default IPs, starting with tier 1 (Shared Pool) before paying for 2 (Datacenter) or 3 (Residential). Set captcha: 1 if a page occasionally shows a simple CAPTCHA wall - it costs no extra credits and is skipped automatically when there is nothing to solve.

File Extraction and Screenshot

# Parse a PDF into Markdown
curl -X POST https://serpforai.dev/api/v1/url \
  -H "Authorization: Bearer $SERPFORAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"s":"https://example.com/whitepaper.pdf","file":1}'

# Full-page screenshot
curl -X POST https://serpforai.dev/api/v1/url \
  -H "Authorization: Bearer $SERPFORAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"s":"https://example.com","image":2}'