Most x402 APIs ship one endpoint. We shipped ten — every one shaped around a real-world agent task.
The URL Metadata API exposes a single line of paid HTTP: GET https://epson-rpm-america-satisfy.trycloudflare.com/api/<route>?url=<URL> returns JSON, paid in USDC on Base mainnet via x402. Settlement address: 0xCa0a6c6Aa7A8F0D5893636CF166Ea2b44fb6500c. No API key. No account. No signup. Pay-per-call.
What it does
Each endpoint specializes in one parsing task so you pay for the payload you actually use:
| Endpoint | Price | Use case |
|---|---|---|
/api/extract |
$0.005 | Full metadata bundle (title, description, OG, headings, links, images) |
/api/summarize |
$0.005 | Title + description + first paragraphs as a text summary |
/api/keywords |
$0.002 | Title + description + keyword list (cheap tier) |
/api/links |
$0.002 | Every outbound <a href> with internal/external split and rel |
/api/markdown |
$0.002 | Clean Markdown extraction (strips nav/ads/scripts) |
/api/og |
$0.001 | Open Graph + Twitter Card tags (smallest payload) |
/api/feed |
$0.001 | RSS / Atom / JSON feed discovery |
/api/headings |
$0.001 | Page outline (every h1/h2/h3 with text + totals) |
/api/canonical |
$0.0005 | rel=canonical + hreflang alternates + og:url + redirect chain |
/api/robots |
$0.0005 | robots.txt policy summary |
A free tier is also available at /api?url=<URL> for anyone who doesn't need the per-call SLA.
Why multiple endpoints instead of one
A typical agent task is asymmetric. A link-preview agent needs og:title + og:image and nothing else — a few hundred bytes. A summarization agent needs the first 3 paragraphs of body text. A SEO crawler needs the canonical URL and every hreflang alternate. Bundling all of that into one endpoint at $0.005 forces every caller to pay for the full parse even when they only use a fraction.
Splitting it into ten routes priced by payload size means a sub-agent workflow can pick the cheapest tier that does the job. A complete crawl of a 10-URL sitemap through the cheapest tier costs 10 × $0.0005 = $0.005 — same as a single call to /api/extract.
What x402 is
x402 is the open standard that turns HTTP 402 "Payment Required" from a joke into a real protocol. The client sends X-PAYMENT with an EIP-3009 transferWithAuthorization signature for the exact amount in USDC. The server verifies and returns the resource. No accounts, no monthly invoices, no Stripe Connect setup — just per-call settlement in ~2 seconds on Base mainnet.
How to use it
from x402 import x402ClientSync
from x402.mechanisms.evm.exact import ExactEvmScheme
client = x402ClientSync()
client.register("eip155:8453", ExactEvmScheme(signer=my_signer))
# Cheapest tier: just the canonical URL + hreflang alternates
resp = client.get("https://epson-rpm-america-satisfy.trycloudflare.com/api/canonical?url=https://example.com")
print(resp.json())
Discovery: GET /.well-known/x402 (legacy) or GET /openapi.json (x402scan canonical).
Built for agents, priced for agents
Total addressable workflows: link-preview generation, summary generation, sitemap enumeration, backlink auditing, content ingestion, SEO audits, multi-region content routing, canonical-URL dedup, feed harvesting, Markdown extraction for context windows. Each route is shaped around one of those tasks and priced accordingly.
Try it free at https://epson-rpm-america-satisfy.trycloudflare.com/api?url=https://example.com or read the LLM-friendly discovery doc at https://epson-rpm-america-satisfy.trycloudflare.com/llms.txt.
(First published 2026-09-09.)
Top comments (1)
The approach of providing multiple endpoints for specific parsing tasks is a brilliant way to optimize both cost and efficiency for users. By allowing customers to pay only for what they need, you’re effectively addressing the asymmetric nature of many agent tasks. A potential improvement could be adding a caching mechanism for frequently requested URLs, which could significantly reduce costs and improve response times. If you're considering expanding the API's capabilities or need additional engineering support for scaling, I’d be open to discussing a paid collaboration. How are you planning to gather user feedback for future endpoint optimizations?