llms.txt — documentation index for LLMs and agentsllms-full.txt — full documentation corpusMarkdown version of the documentationPaddock MCP server descriptor
— Methodology

How Paddock measures agent commerce.

Every figure Paddock publishes is traceable to a documented source and scoped to a stated coverage tier. This page is the reference for citing Paddock data in research, diligence, and analysis.

01 / SOURCES
01

Data sources

Paddock merges independent public sources into a single, cleaned view, and traces every figure to the source that underpins it.

Settlement · Universe 1

x402scan all-facilitator aggregate

The stats.overall settlement aggregate across all x402 facilitators. Basis for Agent Settlement Volume and all transaction counts.

Catalog · Universe 2

CDP Bazaar catalog

Per-origin service listings from the Coinbase x402 Bazaar, refreshed across all pages and chains. Basis for origin-level attribution.

Registry · Universe 3

Registry harvest

A crawl of public agent-service listings, capturing each project's self-declared payTo address and category from its live 402 response or .well-known/x402.json.

Liveness

Liveness probes

Probes of whether listed services actually respond, so attribution and Spend Share reflect live services rather than dead listings.

02 / TIERS
02

Coverage tiers

Agent-commerce data exists at three levels of resolution. Each figure Paddock publishes is drawn from one of them.

UNIVERSE 1
100%
of settled
Agent Settlement Volume
Everything Paddock observes settling on the x402 rail — a census, not a sample, countable in full without attribution. The flagship metric (§03) is drawn from this tier.
Basis for: market-wide totals
UNIVERSE 2
~82%
of settled
Attributed Volume
Settlement traceable to a named origin via the Bazaar catalog. Larger, machine-resolved, but not wallet-verified.
Basis for: category and provider analysis
UNIVERSE 3
~5%
of settled · growing
Verified Mapped Market
Settlement resolved to specific, verified wallets. The smallest and most rigorous tier, expanding through registry harvest and ongoing mapping.
Basis for: Spend Share · health signals
03 / ASV
03

Agent Settlement Volume

Paddock's flagship metric — the headline measure of how much AI agents settle on-chain.

Agent Settlement Volume (ASV)

The total USDC settled through x402 agent payments that Paddock observes via x402scan's all-facilitator aggregate, reported as a 7-day rolling average.

A census-level measure of total indexed x402 settlement, drawn from Universe 1 — no attribution required to state it.

The 7-day rolling average is the headline figure; the raw daily value is reported alongside it as context, because daily settlement is volatile and a single day is not representative of the trend.

04 / ATTRIBUTION
04

Attribution

Attribution maps a settlement wallet to the named service behind it — turning anonymous on-chain payments into category and provider intelligence.

How the map is built

The wallet-to-service map is built from the full Bazaar catalog — all pages, all chains, every payTo — and extended through verified manual mapping. Resolution works forward: it begins from a project's own live payment configuration, not inferred from wallet behavior. Each mapping is checked against the seller's self-declared config before it enters the map.

Excluded wallets

  • Facilitator and pooled wallets are classified and excluded — they carry pass-through volume, not the demand of a single service.
  • Identified circular clusters are excluded from attribution (see §05).
05 / CIRCULAR
05

Circular detection

Not all settlement is demand. Paddock detects circular (self-funded) settlement and reports it as a scoped finding.

Criteria

A cluster is flagged as circular on the basis of self-funding analysis — the share of a cluster's inflow originating from within the cluster itself — corroborated by on-chain reconstruction of the payer side of each settlement.

Scope and scale

The largest identified circular cluster represents a large majority of a single facilitator's volume but only a fraction of a percent of total settled transactions ecosystem-wide. Both are true; they differ only in denominator — which is exactly why the two are never conflated.

Scope limitation

Circular detection currently covers facilitators with on-chain-readable settlement. Clusters on facilitators without a readable settlement path are not yet detectable, and Paddock states this limit rather than implying market-wide coverage.

06 / RAILS
06

Beyond x402

Agent commerce settles across many rails, and they are not equally observable. Paddock measures each to the depth its data allows — and labels which is which.

On-chain settlementFully observable
x402 on Base and Solana. Settlement is publicly verifiable on-chain, so volume, endpoints, and attribution are directly measured. This is Paddock's core dataset.
Directory-indexedPartially observable
Rails such as MPP publish a registry of services but not settlement. Paddock tracks the registered-service count and its change over time — a supply signal, not a settlement figure.
Developer adoptionProxy signal
For emerging rails, Paddock tracks public developer-adoption proxies — repository activity and integration counts — labeled as engagement, not settlement.
Disclosed-onlyNot settlement-readable
Card-network and off-chain agent schemes publish no on-chain settlement. Paddock records what these rails disclose publicly, sourced and dated, without implying a measured volume.
— Keep going

What to read next.

Monthly intelligence
Read the report.

The State of Agent Commerce — the citable monthly report on what AI agents actually buy.

See the report →
x402 market
What agents pay for.

Real transaction volume by category. Three sources merged into one live table.

See the market →