Dataset licence
For research groups, market makers, and institutional buyers who want the underlying data behind our reports and paid endpoints, not the per-agent viewer.
We store only your email and message on this server so we can reply. No account required, no follow-up sequence, no third-party sharing.
What is licensed
Four rolling datasets, extracted from the same store that powers the on-site paid endpoints and the quarterly report. Nobody else joins the paid-endpoint cohort to on-chain ERC-8004 identity. That join is the differentiator; the rest is honest measurement layered on top of it.
x402-earning-endpoint ↔ ERC-8004 identity join
- Fields: resolved endpoint host, publisher registration (chain, agentId, tokenURI), on-chain owner, x402 payment mode, observed capabilities, first-seen / last-seen timestamps.
- Coverage: agents where a paid HTTP surface has been resolved AND matched to an ERC-8004 identity record.
- Refresh: weekly rolling extract.
- Provenance: each row cites the on-chain block that carried the registration and the HTTP probe timestamp that confirmed the endpoint. Nothing inferred, nothing scraped from third-party aggregators.
Solana registry census
- Fields: program-derived agent account, authority, metadata pointer, resolvability outcome, resolvability failure reason (when applicable).
- Coverage: the Solana-side ERC-8004-analogue registry the site indexes.
- Refresh: weekly rolling extract.
- Provenance: direct Solana RPC read; per-account slot recorded so replays are reproducible.
tokenURI resolvability (multi-chain)
- Fields: chain, agentId, tokenURI as registered, resolution outcome (ok / http error / ipfs miss / schema fail / empty payload), resolved MIME type, response byte count, probe timestamp.
- Coverage: every ERC-8004 IdentityRegistry entry on the chains the site indexes (Base, Ethereum, BNB) whose tokenURI has been probed at least once.
- Refresh: weekly rolling extract.
- Provenance: the same probe that drives the on-site “populated tokenURI ≠ resolvable agent card” column. Failure reasons are recorded verbatim — no coercion to a “works / doesn’t” boolean.
Ethereum IdentityRegistry census
- Fields: agentId, owner, tokenURI, registration block, registration transaction hash, deregistration state (when applicable).
- Coverage: the Ethereum mainnet IdentityRegistry.
- Refresh: weekly rolling extract.
- Provenance: direct RPC log read; block-of-record recorded per row so backfills are re-derivable.
Honest-measurement disclosures
These disclosures ship on the on-site surfaces already and travel with every extract — the framing IS the product:
- Per-publisher dedup is capped, and the cap is stated on every extract. Uncapped cohort counts are surfaced separately so consumers can pick the definition they want.
- Registrations ≠ agents. A registration is a row in an IdentityRegistry; whether it corresponds to a live agent is a downstream question the resolvability data answers.
- Populated tokenURI ≠ resolvable agent card. A URI field with bytes in it is not the same as a document a client can fetch and parse. The tokenURI-resolvability dataset carries the reason a probe failed, never coerced to a boolean.
Delivery
- Refreshed extract in a documented schema (JSONL or Parquet, your choice).
- Cadence: weekly rolling. Faster cadences are available on higher scope tiers — we’ll state exactly what we can honour, not what would sound good.
- Change-log per extract: rows added, rows removed, fields changed. No silent schema drift.
Price
From $99/mo, scoped by dataset and refresh cadence. We’ll quote against your actual scope before you commit — no self-serve card form, no auto-renewing sticker.