Samples
Sample responses
Every sample is computed at build time from a dated reading. Nothing here is typed. A field a reading cannot produce is listed under _sample.fields_pending, never invented.
Unpaid GET returns the preview inside the 402; pay to get full. Each paid route lists both: <id>.json is the full response, <id>.preview.json is the 402 body. The preview holds the endpoint's headline aggregates, exact as measured — fewer fields, never softer numbers.
- basket.json · basket.preview.json — Market-level only: Reading #2 holds NO per-seller co-purchase edges. The share is over the 400 wallets in the reading's scatter, a subset of its 1817 payers.
- edge.json · edge.preview.json — Live route, no dated fact yet: the daily cron publishes the first complete UTC day (2026-09-03) at 01:25 UTC on 2026-09-04. Counts only — never ids or addresses.
- labels-infra.json — Status designed: the infra-caller filter is not yet published as a route, so this sample carries no fact.
- seller.json · seller.preview.json — Reading #2 aggregates by payTo wallet and labels shared wallets '+N hosts'; this row is a single-host wallet.
- settlement.json · settlement.preview.json — Block timestamp was not captured by the control. seller_listing carries listed_price_match but not the hosts array.
- settlers.json — 22 facilitators, 84 Base settler addresses from the registry snapshot. The residual is 10 unattributed settler addresses observed by the settler-frame scan in 2026-08-31-settler-frame, as [address, txs] pairs.
- stats.json — Market pulse: the newest daily settler-frame census, headline aggregates. Market totals are the free lead-in and are never gated; entity-level facts are the paid routes. Dollar totals in this frame carry gasless transfers that are not x402 — purchase counts are the honest headline.
- wallet.json · wallet.preview.json — address_prefix is a TRUNCATED identifier: Reading #2 publishes an 8-hex-digit prefix, not the full address. The live route keys on the full address.
- window.json · window.preview.json — Fact taken verbatim from the settler-frame census; coverage is computed from that scan's persisted chunk manifest.