Reliability scores for autonomous agents.
Crypterian observes on-chain activity for ERC-4337 smart accounts on Base and agent-controlled wallets on Solana, then issues each wallet a 0–100 reliability score computed from observed on-chain behavior. Operations are paid per call over the x402 payment standard — USDC on Base or Solana, no API keys, no contracts.
We watch ERC-4337 user operations on Base and program-call activity on Solana every 5 minutes. Every transaction, paymaster, target contract, and outcome is captured. Auto-discovery surfaces new wallets without manual seeding.
Success rate, recency, throughput, paymaster quality, and behavioral fingerprint
combine into a 0–100 score with a confidence interval. Wallets with fewer
than 100 observed events return tier: unranked — we don't
fabricate signal without evidence.
Two endpoints serve the score: a basic call and an enriched call with tier-percentile rank, 90-day callData fingerprint, and sybil-cluster detection. Settlement happens in USDC on Base or Solana, paid per request.
Reliability score, tier (S/A/B/C/D), behavioral category, sybil flag, confidence interval, and observation counts. Auto-detects chain from wallet format (Base 0x… or Solana base58). Designed for runtime scoring decisions.
Everything in /score, plus tier-percentile rank ("which S-tier is best?"),
90-day behavioral fingerprint from decoded callData (distinct contracts, paymasters,
top selectors with labels), and reserved 30/90-day reliability slots. Designed for
onboarding decisions.