In the week of July 6, 2026, of the 109,494 top web domains a Cloudflare Radar agent-readiness scan successfully crawled, only 3 serve x402 payments and exactly 1 advertises an AP2 agent-payment endpoint — the two most-hyped agent-commerce standards are, in raw web adoption, effectively at zero. The broadest signal, UCP, matches 8,099 domains (a suspected over-detection), while 246 publish an MCP server card and 27 an A2A agent card.
Domains serving each agent standard (raw counts — shares not comparable) · domains
Domains serving each agent standard (raw counts — shares not comparable), in domains
Period
domains
UCP
8,099
MCP server card
246
web-bot-auth
43
A2A agent card
27
ACP
10
MPP
7
x402
3
AP2
1
Adoption here is a weekly scan, not a live counter
These figures come from a weekly agent-readiness scan of the top web domains, so the as-of stamp on this page is a data week, not a single day. A value that ticks up between Monday and Thursday is the same weekly measurement re-read; treating it as a daily series over-reads noise that is not there. Month-over-month change compares this week's scan against the scan from roughly a month earlier — the honest grain for a signal that moves slowly.
Why these shares cannot be compared to each other
Each check has a different meaningful denominator, so ranking standards by their bare share across all domains is a category error. x402 is a payment header for paid content and APIs — it only belongs on the small slice of sites that actually sell access — so its share against every crawled domain understates its penetration of the sites where it would ever appear. AP2, MPP and ACP are similarly niche. A crawl-control file that belongs on every website and a payment endpoint meant for a few thousand commerce sites are not on the same axis. The counts below are raw domain matches; the shares are shown for scale, never for cross-standard ranking.
The UCP number to distrust
UCP's count sits an order of magnitude above every other agent standard, and that gap is most likely a detection artifact rather than real adoption. The scan's UCP check reads as a loose heuristic that over-matches domains which do not actually implement the Universal Commerce Protocol. Treat the UCP row as a suspected over-count, and weight the smaller, stricter checks — MCP server cards, A2A agent cards, x402 — more heavily when reading genuine adoption.
Also asked
Does a low x402 domain count mean x402 is failing?
No — x402 is a payment standard for paid content and APIs, so it only belongs on the small subset of sites that sell access. Its share across all crawled domains understates its penetration of the sites where it would ever appear, and on-chain x402 settlement counts (millions of transactions) tell a very different story from web-domain presence.
Why is the UCP number so much higher than the others?
It is almost certainly a detection artifact. The scan's UCP check reads as a loose heuristic that over-matches domains, so the UCP row should be treated as a suspected over-count rather than real adoption of the Universal Commerce Protocol.