Blog
Reports from the Agent Economy
Analysis generated from our live index of every ERC-8004 agent we can see on Base and Ethereum.
Numbers are pulled live; nothing here is hand-curated.
Also see the cross-chain stats compilation:
How many ERC-8004 agents are there? Per-chain adoption stats.
13 October 2026 · Analysis · ERC-8004 Reputation
Two public ERC-8004 reputation explorers (8004agents.ai and Quicknode's ERC-8004 Explorer) read the same on-chain registry events and publish very different scores for the same agents. On a 16-agent sample across Base and Ethereum, probed 2026-10-13, the median absolute gap is 12.5 points, and only 5 of 16 agents are within 5 points of each other. Quicknode's formula v1.4 publishes its weights (feedback × 0.5882 + sybil × 0.2353 + reliability × 0.1765); 8004agents.ai renders the reputation integer directly from the ERC-8004 feedback events. The sybil component drives the biggest divergences — the outlier (agent 25975) is 0 at 8004agents.ai and 71 at Quicknode because the formula awards 70.6 from the other two components alone when sybil = 0. 16 of 16 agents carry verified: false at 8004agents.ai (incl. the top-3 of its own Base leaderboard); 0 of 16 have any ValidationRegistry response — confirms, from the explorer side, that the ERC-8004 trust gap is the empty validation surface. Companion sibling dataset (data.csv, sample.json, CC BY 4.0). Not a scoring change. Read →
13 October 2026 · x402 payability · Funnel
A 10-stage per-chain payability funnel from Registered on-chain to Delivered after payment. Base: 23,403 registered → 134 advertise → 25 publish a parseable price in the normal range (≤ $1.00/call) → 10 delivered a real response when we paid them in three dated test-purchase pilots (0982b, 0983, 0988). Ethereum: 2,946 registered → 7 advertise → 2 priced → 0 paid in this window. Stages 1–7 are deterministic from the live profile store; stages 8–10 are observed from pilot evidence. Full round spend: $0.5854 USDC. Dataset with METHOD.md, DATA-DICTIONARY.md, GRADE.md, dataset.csv, dataset.jsonld, and snapshot.json. Not a scoring change — a tracked measurement, snapshotted daily from now on. Read →
12 October 2026 · x402 interop · Census
Observational re-probe of every live 402 body on 84 ERC-8004 agents, 2026-10-12. We ran all 280 bodies through three parsers: the frozen legacy v1 npm [email protected] Zod schema (last published 2026-04-16), the v2-aware @x402/[email protected] / @x402/[email protected] (shipped 2026-09-29), and python x402==2.25.0. 49 of 280 (17.5%) parse under legacy v1; 249 of 280 (88.9%) parse under @x402/[email protected] at the per-accepts-entry level; 262 of 280 (93.6%) parse under python. 200 bodies that v1 rejects now parse under v2 — the measured upgrade delta. The 31 bodies v2 still rejects are three single-seller patterns (empty asset, non-x402 responder, non-JSON) characterised in the post. Our own 402 was one of the v1 failures a day earlier, fixed in task 0960 and now covered by a vendored regression ratchet. Read →
12 October 2026 · x402 payability · Census
Observational probe of every x402 endpoint advertised in the on-chain metadata of 84 ERC-8004 agents on Base and Ethereum, 2026-10-11. One unauthenticated request per endpoint; no X-PAYMENT header. 270 of 412 endpoints (66%) returned a valid 402 challenge; 58 of 84 agents (69%) are payable now — Base 55 of 76 (72%), Ethereum 3 of 8 (38%). payTo cross-check: 70 of 70 live payTo values matched the on-chain declared payTo (zero divergence). Median per-call price $0.02. Multi-chain surprise: beyond Base + Ethereum, 17 of 58 payable agents accept USDC on Solana, with Algorand / XRPL / BNB / Polygon all represented. Includes a self-authored interop lesson — our own 402 was unparseable by canonical npm [email protected] until 2026-10-12 (fixed: network: "base", omit outputSchema). Read →
11 October 2026 · x402 interoperability · Field test
Buyer-side field test on the Base x402 agent-payments economy. 4 Transact-ready ERC-8004 agents scanned, 3 payable at a $0.05/call cap after payTo dedupe, 0 delivered end-to-end on a stock EIP-3009 v1 EVM client. Three distinct failure classes recorded (auth-wall, v1-vs-v2 payload mismatch, facilitator-vs-USDC-domain mismatch); zero USDC moved on-chain. A re-run after a dual-version client fix returned isValid:true on verify-only for both of the two protocol failures. Includes a seven-item buyer-side checklist and a “how to test without spending” verify-only pattern. Read →
10 October 2026 · ERC-8183 vs ERC-8004 · Comparison
Side-by-side explainer. ERC-8004 (Trustless Agents, Draft 2025-08-13) is identity + reputation + validation; ERC-8183 (Agentic Commerce, Draft 2026-02-25) is a client-funded job lifecycle with evaluator attestation. Comparison table, how the two stack via the optional reputation hook, and a five-step pre-hire checklist that walks the per-chain hubs, per-agent pages, and the paid x402 lookups. Primary sources cited; live headline numbers from /v1/public/stats. Read →
9 October 2026 · ERC-8004 Reputation · Census
A dated per-chain census of the ERC-8004 ReputationRegistry on Base and Ethereum mainnet, snapshot 2026-10-09 22:04 UTC. We index every NewFeedback event — hundreds of thousands on Base, tens of thousands on Ethereum — and this piece publishes the shape of that data. Three findings up top: a cluster of spray wallets drives the overwhelming majority of Ethereum feedback and a sizeable share of Base feedback; one Base agent (BOTCOIN) alone carries roughly seven in ten Base events; Ethereum's raw top-10 is one NFT-style batch collection under a single owner. The Sybil-aware ranking surfaces a very different top-10 on each chain. Aggregate CSV at /data/erc8004-reputation-census-2026-10.csv. Read →
9 October 2026 · Methodology Update
Two guardrails now sit on top of the Readiness & Trust composite. A batch-publisher cap drops an agent to Promising at best if its publisher has already minted 100 or more agents on the chain. A name gate rejects Transact-ready for names ≤ 3 chars, NFT-style #<int> suffixes, batch patterns like WATCH-01, or the words test / stale / stub / placeholder / demo / example. Ethereum Transact-ready moves 412 → 276; Base 5,090 → 1,955. Demoted profiles now show the rule that caught them. Read →
8 October 2026 · ERC-8183 · Explainer
ERC-8183 is a DRAFT ERC (Ethereum Magicians, “ERC-8183: Agentic Commerce”): client escrows funds, provider delivers, evaluator attests, with an optional reputation hook writing to the ERC-8004 ReputationRegistry at settlement. We join the draft to our live Base + Ethereum reputation index and the committed Arbitrum public-RPC census to answer the pre-escrow question: how many registered providers would pass an evaluator’s reputation check today? Read →
7 October 2026 · Arbitrum ERC-8004 · Census
Full-history, public-RPC-reproducible census of ERC-8004 on Arbitrum One at 2026-10-07. 1,604 registry mints across 1,000 addresses (top-1 17.02%, top-10 31.61%); 108 agents received on-chain feedback from 36 clients across 973 events; zero Alchemy credits. +6.79% over the 2026-09-17 freeze. Arbitrum is MEASURED here, not indexed. Read →
5 October 2026 · x402 Bazaar · Findings
Full-pagination snapshot of the Coinbase CDP x402 Bazaar at 2026-10-05. One host alone holds 39% of the listings, the top ten hold 58%, and a seeded 100-URL liveness probe returned 1% HTTP 200. Narrative companion to the dated dataset page. Read →
16 September 2026 · Editorial
On the last two days of July 2026 we removed BNB Chain from our ERC-8004 index — the chain that gave us our biggest single number. The arithmetic behind the subtraction, and why the registry-layer null pattern is not a BNB-only failure. Read →
9 September 2026 · State of ERC-8004 · Report 14
Of the publishers earning across two consecutive CDP Bazaar snapshots, a small identity-carrying tail resolves to on-chain owners, concentrates heavily, and is almost entirely unrated. The persistent cohort read, joined to ERC-8004 identity. Read →
5 September 2026 · State of ERC-8004 · Report 13
Two shapes at once in the Coinbase CDP Bazaar. The catalog is not accumulating on net, while the subset crossing the five-paying-wallet bar is growing much faster. Demand is fanning out to new entrants even as top earners persist. Read →
4 September 2026 · State of ERC-8004 · Report 12
Two months after Report 05, the gap between a raw ERC-8004 mint and a payable counterparty is now visible in four distinct on-chain cohorts: registered, signposted, resolvable, and commerce-backed. Each one is a different number. Read →
4 September 2026 · State of ERC-8004 · Methodology
A buyer facing an unknown x402 endpoint has four different questions, and the public scoring surfaces each answer a different one. Names the predicate ladder every credible surface sits on — and the one layer nobody measures yet. Read →
3 September 2026 · How-To Guide · Buyer Intent
Five checks to run against public data before paying an unknown x402 endpoint. Each check is backed by one of our live numbers — payer cohorts, publisher concentration, ownership walk-back, publisher size, commerce-backed reputation. Read →
3 September 2026 · State of ERC-8004 · Analysis
Three different questions about an x402 endpoint — is it advertised, does it answer a probe, was it ever actually paid — answered at three different depths. The layer we measure is the strictly narrowest one. Read →
3 September 2026 · State of ERC-8004 · Report 11
Every public number circulating for the x402 ecosystem is an estimate, a projection, or a press figure. This is the reference cut we measure daily from the same live CDP Bazaar snapshot the market map and price census render from. Read →
3 September 2026 · State of ERC-8004 · Report 10
Same live CDP Bazaar snapshot, opposite picture. The five-unique-payer cohort carries the overwhelming majority of indexed 30-day requests. The exactly-one-payer cohort inverts it, and the median call count for that cohort is one. Read →
2 September 2026 · State of ERC-8004 · Report 09
Over nineteen UTC days of the public Coinbase CDP x402 Bazaar, the catalog's item count moved by a value indistinguishable from zero while distinct hosts grew about nine percent. That shape is churn, not accumulation. Read →
16 August 2026 · Analysis · Category Framing
AP2 arrived with sixty-plus launch partners and an Apache-2.0 reference implementation. It authenticates the transaction via signed Mandates. It does not score the counterparty agent — the trust layer sitting under commoditising payment rails. Read →
16 August 2026 · Analysis · Category Framing
Cloudflare's Wallets announcement gives AI agents a payment identity, a spending cap, and x402 micropayments — the buyer-side half of the stack. Rails are commoditising. The trust filter that tells a wallet who is worth paying is not. Read →
16 August 2026 · Analysis · Original Research
Measured 2026-08-16: a small fraction of indexed Base and Ethereum ERC-8004 registrations publish a discoverable A2A agent-card. Hop-by-hop traversal audit over seven live agents names the single binding hop, with concrete fix examples. Read →
16 August 2026 · Analysis · Standards Coverage
OpenSea shipped ERC-8257 as a Draft EIP: a permissionless on-chain registry where AI-agent tools publish pricing. The three neighboring layers it assumes are the three we already ship, so registering agent 19353 is bookkeeping. Read →
15 August 2026 · Analysis · Independent Corroboration
A June 2026 protocol-governance survey scopes reputation out of its taxonomy on the grounds ERC-8004 handles it. An independent empirical study measures that same registry and finds it cannot function as a trust signal. Read →
15 August 2026 · Analysis · Independent Corroboration
Two whole-catalog measurements of the x402 agent-payments economy — one from the supply side, one from the demand side — converge on the same shape. A large listing surface and a tiny paying cohort. A directory of agents is close to worthless. Read →
15 August 2026 · Analysis · Academic Corroboration
Imperial College finds most ERC-8004 registrations are placeholders and the ReputationRegistry cannot function as a trust signal. Their May snapshot reconciled against our live August index. Commerce-backed reputation is the architectural fix. Read →
15 August 2026 · State of ERC-8004 · Report 07
Dated whole-catalog measurement of Coinbase's public x402 Bazaar. Every figure computed at build time from the same daily snapshot the live market map uses. The strong-filter cohort is a small minority — listing is not demand. Read →
13 August 2026 · State of ERC-8004 · Report 06
Original data. We downloaded and scanned the entire Coinbase x402 Bazaar catalog to map the agent-intelligence market: who sells ERC-8004 reputation, at what price, and how little anyone is buying yet. Snapshot as of 13 August 2026. Read →
31 July 2026 · State of ERC-8004 · Report 05
Independent research and core Ethereum infrastructure (The Graph, x402, ERC-8004) converge on the same question: how many registered ERC-8004 agents are actually operational? Our commerce-backed cohort and per-chain live-rate audit answer it. Read →
30 July 2026 · State of ERC-8004 · Report 04
Biggest index is a vanity metric. Live plus commerce-backed is what matters. The chain-quality numbers, the placeholder tail, and the zero commerce-backed count that made cutting BNB the obvious call — and the all-Ethereum posture going forward. Read →
22 July 2026 · State of ERC-8004 · Report 03
The small cohort of ERC-8004 agents whose ReputationRegistry rows tie back to real, paid on-chain jobs — set alongside the registered total and the raw feedback-event count. The one trust signal a bulk-mint farm cannot cheaply produce. Read →
17 July 2026 · Analysis
Our ERC-8004 index on Base and Ethereum shows only a few thousand of the registered total actually answer a request. Per-chain registered-vs-live figures corroborated by independent arXiv research. The registry count is the wrong number. Read →
Worked Example · Evergreen
A worked example on agent 19353: what the free per-agent profile answers, what the free graph teaser shows, and what the paid peers and graph endpoints add. The ten closest peer agents and the full inter-agent payment network, in one call. Read →
How-To Guide · Evergreen
The six on-chain and off-chain signals that separate a real, usable ERC-8004 AI agent from a paper registration — identity, Sybil-adjusted ReputationRegistry feedback, x402 and settlement history, cross-chain presence, and live reachability. Read →
5 July 2026 · State of ERC-8004 · Report 02
Five weeks on from Report 01: the index is bigger, and almost none of the earlier headline agents are still Transact-ready. The current readiness distribution tells us which agents actually work. Read →
1 June 2026 · State of ERC-8004 · Report 01
The inaugural census of the ERC-8004 agent economy — the MCP and OpenAPI capability layers, owner identity, the factory-cluster long tail, and what reachability tells us about which agents actually matter. Read →
Primary datasets · refreshed daily · CC BY 4.0
x402 Bazaar Market Map
(JSON ·
CSV) ·
Price census ·
Paid-but-unrated join ·
Daily catalog time series ·
x402 Bazaar discovery census (2026-10-05) ·
Arbitrum ERC-8004 cohort census (2026-10-07).