Summary
Scoring the reviewers rather than only the agent is what makes the rating hard to game.
Any reputation system built on feedback gets farmed by fake reviewers. RNWY treats the reviewer set as the thing to audit — wallet ages, velocity, sweep patterns, clustered scores — so a perfect rating assembled from week-old wallets reads as what it is. Wallet age cannot be backdated, which is why they lean on it. Take the attestation signature seriously too: a verdict you can verify later is different from one you have to trust now.
What it is
RNWY grades AI agents registered on-chain, across the ERC-8004, Olas and Virtuals registries. It answered our probe anonymously.
What you get
- trust_check — a pass or fail verdict for an agent, with score, tier and a signed attestation
- reviewer_analysis and reviewer_wallet — wallet ages and behaviour profiles of the reviewers, detecting sybil patterns such as sweeps and score clustering
- address_age — wallet age in days, described in their own tool text as the uncheatable signal
- compare_agents, entity for an operator's full footprint, commerce_stats and network_stats
- mcp_attestation — a signed risk assessment for an MCP server, and risk_terms for counterparty risk tiers
Requirements
Nothing to start. Add the URL to any client and it works anonymously; a free API key raises the rate limit.
Setup effort
Paste a URL — add the endpoint to your client