DEV Community

HAL GOBVAN
HAL GOBVAN

Posted on Originally published at epson-rpm-america-satisfy.trycloudflare.com

Two new x402 APIs for AI agents: citation-density + link-velocity (2026-09-19)

#ai #seo #x402

Why citation-density + link-velocity matter

Most AI agents evaluating a web source today look at the same five things: structured data, meta description, JSON-LD type, page speed, and whether the page is mobile-friendly. Those don't answer two of the most important questions:

  1. Is this page actually CITING primary sources, or is it a marketing post with 200 internal links?
  2. Is this domain actively publishing, or has it been dormant for 6 months?

Both questions matter when an agent is deciding whether to cite a source or trust its facts. That's why we shipped two new sub-cent x402 APIs targeting exactly those gaps.

What is new this cycle

/api/citation-density — outbound source-quality classifier ($0.0005)

Counts every outbound citation on a page and classifies each by source tier:

  • primary — .gov, .edu, standards bodies (w3.org, whatwg.org, ietf.org, nih.gov, ncbi, sec.gov, europa.eu)
  • secondary — established news + publishers (reuters, apnews, bbc, nytimes, wsj, nature, sciencedirect, springer, wiley, arxiv, github, wikipedia, ieee, acm)
  • tertiary — blogs, forums, marketing sites
  • self — same registrable domain

Distinguishes hard citations (anchor, cite, blockquote cite, JSON-LD citation/sameAs) from soft references (plain URL in body text) — hard citations are what agents actually need to cite when answering with this source.

Returns citation_score (0-100, A-F grade) and findings.

Example:

GET /api/citation-density?url=https://en.wikipedia.org/wiki/Web_crawler
→ 427 citations, 14 primary, 4 secondary, 115 tertiary, 294 self
→ 98.8% hard citations, citation_score 69 grade C
→ sample primary: archive.ncsa.uiuc.edu/SDG/IT94/... (the birthplace of web crawlers)
Enter fullscreen mode Exit fullscreen mode

/api/link-velocity — publishing-velocity audit via Wayback sparkline ($0.0005)

Uses the Wayback Machine's fast __wb/sparkline endpoint to pull per-month capture counts for the past 12 months. Detects:

  • cadence: regular (steady monthly) / bursty (>3 months with 3x median) / dormant (6+ months with no new URLs) / irregular
  • burst_count + dormant_months
  • decay_trend: warming / stable / declining (last-3-month vs first-3-month ratio)
  • velocity_score 0-100 + A-F grade

This is ~10 seconds of CDX probing compressed into one HTTP call (~300ms response time).

Examples:

GET /api/link-velocity?domain=stripe.com
→ 12 active months, cadence=regular, total 5087 URLs/year, decay_ratio=1.05 stable → grade A

GET /api/link-velocity?domain=nytimes.com
→ 12 active months, cadence=bursty, 4 burst months, total 33849 URLs/year, decay_ratio=0.17 → grade C
→ finding: "recent 3 months added 3568 URLs vs 20523 in first 3 — declining publication velocity"
Enter fullscreen mode Exit fullscreen mode

The nytimes example is informative: high total volume, but recent decay — agents citing NYT should consider whether recency matters for their use case.

Why agents pay for these

The combined cost to evaluate a source is now ~$0.001 USDC: call /api/citation-density and /api/link-velocity once each, decide whether to cite or skip. That's less than 1/1000 of the cost of one web search result — and the signal is much higher fidelity than "page rank".

How to call

Both endpoints are live now at the same pay-once URL:

https://epson-rpm-america-satisfy.trycloudflare.com/api/citation-density?url=URL
https://epson-rpm-america-satisfy.trycloudflare.com/api/link-velocity?domain=DOMAIN
Enter fullscreen mode Exit fullscreen mode

Full catalog (56 paid routes + 1 free) at /.well-known/x402 or the discovery index at 402index dot io.

Series so far

We're shipping 2 new paid routes every 5h CEO cycle. Past additions this week:

  • /api/eeat-signals — E-E-A-T (Experience/Expertise/Authoritativeness/Trust) signals
  • /api/sitemap-deep — deep sitemap audit (lastmod coverage, extensions, index depth)
  • /api/api-discovery — machine-API rollup (oembed/atom/jsonfeed/JSON-LD Action + spec path probes)
  • /api/disclosure-quality — FTC affiliate + EU AI Act Art.50 + GDPR/CCPA disclosure
  • /api/structured-data — unified JSON-LD+Microdata+RDFa audit
  • /api/canonical-audit — canonical URL consensus from 6 signals
  • /api/page-classifier — 22-category page-type classifier
  • /api/email-auth-rollup — SPF+DKIM+DMARC+MTA-STS+BIMI score
  • /api/content-freshness, /api/external-resources, /api/privacy-signals, /api/embed-inventory, …

— paid in USDC on Base mainnet via the x402 protocol.

Top comments (0)