--- title: "The corpus" description: "Weekly signed observations of the x402 ecosystem — which listed hosts answered and what a conformance probe saw. Hash-chained, ed25519-signed." canonical: "https://scvd.store/corpus" url: "https://scvd.store/corpus" data: "https://scvd.store/corpus.json" --- # The corpus Weekly signed observations of the x402 ecosystem — which listed hosts answered and what a conformance probe saw. Hash-chained, ed25519-signed. ## What this is Weekly signed observations of the x402 ecosystem: which listed hosts answered, and what one conformance probe saw at that moment. Hash-chained, ed25519-signed, each digest submitted to OpenTimestamps for Bitcoin anchoring. Free to read. ## Census A census of every host on the public x402 discovery list (2026-08-03, one GET each, reproducible by anyone via the free checker at POST /api/preflight/v1) found 34 of 35 hosts serve no signed offers in the challenge header that probe reads — offers placed only in the 402 body were beyond that instrument's read — and the one attempting them serves offers that fail JWS parsing before a verifier reads a single field. Our own door, which serves signed offers, is not in the 35: a census cannot probe its own host, and that exclusion flatters nobody but us, so it is stated. ## Data ## Every door listed ## Per host ## Weekly brief ## Reports ### Withdrawn — The x402 ecosystem, walked with a wallet: August 2026 #### ID x402-ecosystem-2026-08 #### Path /api/report/x402-ecosystem-2026-08 #### Published 2026-08-19 #### Modified 2026-08-20 #### Page modified 2026-10-06 #### Withdrawn ##### At 2026-08-20 ##### Reason The largest failure class was attributed to sellers; this report's own committed ledger supports that reading for about 3% of it. Roughly 29% were endpoints correctly asking for inputs our instrument never sent, and 41% answered with an empty body and cannot be read either way. The headline two-thirds figure therefore measures client-plus-server friction on one hand-rolled instrument, not seller failure, and the report did not say so. ##### What stands The chain-side arithmetic is unaffected and stays true: 669 USDC transfers left the declared wallet totalling $6.396969, against 489 the client recorded as paid — the 180-settlement recording gap is a fact about the instrument's own bookkeeping and, if anything, this correction strengthens it. The raw ledger, the transfer list and the scripts remain committed at research/field-run-2026-08-18/ so anyone can redo this classification and check ours. ##### Next A re-run with the instrument repaired — required inputs sent, chainId read from the offer, accepts filtered to a rail we can actually sign, and success counted on any 2xx — would produce a number worth publishing. Until then the store makes no claim about ecosystem payment-failure rates. #### Description Withdrawn 2026-08-20. The x402 ecosystem, walked with a wallet: August 2026. Original signed report retained with the withdrawal, evidence and verification instructions. #### URL ## As time ## Since diff ## Latest ## Changes by week ## Asked for ## Subscribe ### How Poll https://scvd.store/corpus/latest.json with If-None-Match (the ETag is the sha256 of the bytes) or If-Modified-Since (Last-Modified is the snapshot's taken_at): a 304 costs nothing, a 200 is a new signed week. Then read https://scvd.store/corpus/changes/{week}.json for what moved, as fields and as a changelog — changed_pay_to is the row to read before paying: where a door asks to be paid moved between two signed weeks, as digests. Or subscribe to the Atom feed at https://scvd.store/feeds/corpus.xml, one entry per signed snapshot. ### Example curl -sS -i https://scvd.store/corpus/latest.json -H 'If-None-Match: ""' ### Notebook ## Wallet facts ## Battery delta ## Seller declarations ## Standing notes ## How to verify Printed on the document itself, at https://scvd.store/corpus.json — recompute the digests, check the signatures against the published key, and run ots verify on evidence that is not ours. ## Corrections