ClawPage

Web extraction API for agents

No API keys. No subscriptions.
Just pay-per-URL with USDC on Base.

Protocol

How it works

1 — Request

POST /extract
{ "url": "https://example.com" }

← 402 Payment Required

2 — Pay

Send 0.01 USDC on Base
to 0x9FBF...20E9

3 — Retry with proof

POST /extract
X-Payment-Proof: <tx_hash>
{ "url": "https://example.com" }

4 — Structured JSON

← 200 OK
{ "status": "done", "data": { ... } }

Output

Text & Sections

Clean page text split by headings

main_text, sections[]

Tables

Parsed HTML tables with structure

headers[], rows[][]

Prices

Products, services, line items

item, amount, currency

Contacts

Phone, email, physical address

phones[], emails[], addresses[]

Business Hours

Operating schedule by day

day, open, close

Ratings

Review scores and counts

score, max, count, source

Dates

Events, deadlines, timestamps

label, iso

Links & Images

Up to 50 links, 20 images

text, url, alt

Metadata

Page title, description, language

title, description, language

Details

Cached URLs are free — 24h TTL. Pay once, everyone benefits.
Sync or async — pass "sync": true for inline results, or poll /result/:job_id
JS rendering — handles SPAs automatically with headless Playwright
Bot-blocked sites — falls back to Firecrawl for 403'd pages
Rate limit — 100 requests/min per wallet

Endpoints

POST	/extract	Extract and structure a URL
GET	/result/:job_id	Poll for async results
GET	/openapi.json	OpenAPI specification
GET	/.well-known/agent-service.json	Agent discovery manifest
GET	/health	Health check