A dated, public-RPC-reproducible census of the ERC-8004 IdentityRegistry and ReputationRegistry on Arbitrum One, generated 2026-10-07T18:29:18Z. One snapshot. Full history from the deploy block (428,895,443) to head (512,646,051) — 83,750,609 blocks, 100% coverage, 228 RPC calls, zero Alchemy credits. Published so answer engines and builders have a citable Arbitrum One ERC-8004 cohort read; the paid /v1/intel/* API still indexes Base and Ethereum. This is MEASURED, not INDEXED.
summary.json):.agent names rather than HTTP endpoints. Prior freeze (task 0656, 2026-09-17): 1,502 registrations — +6.79% in ~20 days.
eth_getLogs was issued in nine 10,000,000-block windows from the ERC-8004 deploy block to head. Deploy block was seeded from the committed 0656 freeze because the public node is non-archive — but eth_getLogs is index-based and works globally, and the scan is complete over the full-history range (coverage_blocks = range_blocks = 83,750,609). Owner distribution was decoded from the Transfer(from=0x0, to=OWNER) to-address of each mint. A 200-of-1,604 evenly-spaced-by-agent-id tokenURI sample (cap=200) was fetched, parsed, and probed with a 5-second-timeout single GET on the first HTTP(S) services[].endpoint found in resolved metadata.
Full method — block windows, sample strategy, probe design, archive-vs-full-node notes — is in METHOD.md. The measurement script is at scripts/analysis/0772_arbitrum_public_rpc_census.py in the repo and the input envelope is at .cc-agent/.0772-arbitrum-census.json, SHA-256 2ce26378ad36a7ded274f9f60ccd6b360b0239e574f5bceac39961260ba1554b.
| Contract | Address | Deploy block | Bytecode length @ head (hex chars) |
|---|---|---|---|
| IdentityRegistry | 0x8004A169FB4a3325136EB29fA0ceB6D2e539a432 | 428,895,443 | 260 |
| ReputationRegistry | 0x8004BAa17C55a88189AE136b182e5fdA19dE9b63 | 428,895,443 | 260 |
Both are the deterministic ERC-8004 addresses — the same contract bytes also deployed on Base, Ethereum, and BNB Smart Chain. Deploy-block match against the four-chain 2026-09-17 census: true.
Top-10 owners by mint count, verbatim from summary.json:
| Owner | Count | Share |
|---|---|---|
0xad7d4d39badd41ca55a0c32b08b12696f2eee721 | 273 | 17.02% |
0x103040545ac5031a11e8c03dd11324c7333a13c7 | 133 | 8.29% |
0xbf0e0e7afde9b7f42f5d93d3aebde53af14bba70 | 54 | 3.37% |
0x52ce108ae72712929fd2b8c6177a9b39e0ccf644 | 10 | 0.62% |
0xda977767452c5dd021624511f14df67b6c9c2c1b | 9 | 0.56% |
0xd3f9dcd6011e1aa13eeb277d9ce5f2f7c9bb6070 | 7 | 0.44% |
0x0dae6e01a88826b4e77d717d9639b64f749c0152 | 6 | 0.37% |
0x44a3d4b120f7d4f403e99062934a788c61f1aec6 | 6 | 0.37% |
0x1a8a3373bf1aeb5e1a21015e71541ff4be09ee41 | 5 | 0.31% |
0xa93fc5280d63d11e809b77432a941d64edc0958e | 4 | 0.25% |
Concentration is real but not single-vendor: the long tail (~990 owners sharing ~68% of mints) is a wider operator population than Arbitrum's prior public measurements suggest. The 17.02% top-1 is lower than the BNB cluster-sample reads but higher than Base (where the top-1 is 11.58% per the 2026-09-17 four-chain census). No batch-mint factory dominates — the top-1 holds 273 of 1,604, not 10,000+.
Full-history NewFeedback event scan on the Arbitrum One ReputationRegistry: 973 events total, from 36 distinct client addresses, to 108 distinct feedback-receiving agents. This is a full-cohort full-history count — not a sample. 108 of 1,604 registered agents (6.73%) have at least one on-chain reputation signal attached to their agent id.
For reference, the paid-API chain-stats surface reports agents_with_reputation = 29,128 on Base and 0 on Ethereum (ETH reputation indexer paused per the project memory). These are different cohorts and no ratio claim is made.
Sample: 200 of 1,604 agents, evenly spaced across agent-id space (sample_strategy=evenly_spaced_by_agent_id, sample_cap=200).
| Metric | Count | % | Note |
|---|---|---|---|
| Carries a tokenURI | 198 / 200 | 99.0% | near-universal; the IdentityRegistry stores one |
| Resolved to parseable JSON | 44 / 200 | 22.0% | data:, IPFS, HTTPS metadata hosts |
| Carries service/capability signal | 40 / 200 | 20.0% | services[], capabilities[], x402Support, payment |
| x402Support flag present | 4 / 200 | 2.0% | small-n |
| Payment field present | 0 / 200 | 0.0% | |
| First endpoint returned 2xx/3xx/4xx | 0 / 200 | 0.0% | HTTP reachability — see caveat below |
The 0/200 HTTP-live rate is a HTTP-reachability measurement on the first endpoint of a sampled agent — it is NOT an agent-liveness rate. Many Arbitrum agents publish .eth, .arb, .agent, or ENS-reverse names in their endpoint field rather than HTTP(S) URLs, and the probe does not resolve those name-service references. The stronger live signal on Arbitrum is on-chain reputation feedback (108 distinct agents, above). A proper HTTP liveness rate for the Arbitrum cohort would require full endpoint enumeration and name-service resolution; those are not in scope here.
| Cluster | Count in sample (n=200) |
|---|---|
clawnews.io | 1 |
vitalik.arb | 1 |
z.agent | 1 |
web3smart53.agent | 1 |
sparkmind31.agent | 1 |
| Date | Head block | Registrations | RPC | Note |
|---|---|---|---|---|
| 2026-09-17 (0656) | 505,982,474 | 1,502 | Alchemy archive | prior freeze — see the four-chain census |
| 2026-10-07 (0772, this dataset) | 512,646,051 | 1,604 | Public RPC (non-archive) | +102 mints (+6.79%) over ~20 days |
This dataset is a published MEASUREMENT. Arbitrum One is not currently covered by the paid /v1/intel/* API, which indexes Base and Ethereum. The verdict in the source envelope (ADD_ARBITRUM(live=108)) is an input to a future indexer-wiring decision — a separate, propose-first change — not an advertisement of current paid coverage. For the two chains the paid API does cover, see /docs and /services.
| File | Type |
|---|---|
dataset.csv | text/csv — 25 rows, 4 columns (metric, value, unit, note) |
summary.json | application/json — all headline aggregates + claim traps |
dataset.jsonld | application/ld+json — schema.org Dataset metadata (also inlined above) |
PROVENANCE.json | application/json — envelope sha256, repo HEAD, verdict, guardrails |
METHOD.md | text/markdown — reproduce in one command |
DATA-DICTIONARY.md | text/markdown — column-by-column reader's map |
summary.json.
/v1/intel/* API covers Base and Ethereum. Discovery is at
/.well-known/x402.json; a free read-only probe surface is at /mcp.
Onchain Agent Intel (2026). ERC-8004 Arbitrum One cohort census (public RPC) — 2026-10-07. CC BY 4.0. https://onchainagentintel.io/data/arbitrum-erc-8004-census-2026-10
This dataset is published under Creative Commons Attribution 4.0 International (CC BY 4.0). Free to reuse, remix, and redistribute — including commercially — with attribution.