ERC-8004 gives an agent a score, and Monad’s own docs tell agents to check it before a high-value transaction. Nothing tells you what that score is made of. Headwater indexes every agent on Monad and traces the funding behind every reviewer, so you can see whether a score is broad, thin, self-written, campaign-funded or produced by a game.
Streamed from our Envio HyperIndex endpoint over a live subscription, newest first. The label is the verdict from the published snapshot. An agent rated since the snapshot shows NEW SINCE SNAPSHOT, and Agent Lookup works out its verdict live.
Its sybil check divides unique reviewers by total reviews. Agent 182 has 7,665 reviews from 7,665 different wallets, so it scores perfectly. It does not look at who funded those wallets.
Each reviewer wallet got 11 MON from the owner, rated the agent, then sent about 10.93 MON back to that same owner. ROUND TRIP
Quicknode figures read from erc-8004.quicknode.com/agents/monad-mainnet/182 and its published formula v1.4 on 2026-10-08. Its overall score for agent 182 is 41, because it drops the review values for using an unlisted tag. The sybil part still reads as perfect. Both readings are correct about what they measure. Counting distinct wallets and asking where their money came from are different questions, and only the second one sees a campaign.
Measured live on Monad mainnet, 2026-10-08, across all 10,324 registered agents. None of this is an accusation. A funded campaign can be a legitimate quest, and a game contract reporting wins and losses is the standard working exactly as designed. The point is that all six of these render as the same kind of number in a registry, and nothing today tells them apart. The badges here are the same words the lookup page, the JSON API and the MCP server use, so nothing on this site calls the same thing by two names.