Sentinel - a developer-first trust layer for APIs

Sentinel is a developer-first trust layer for APIs. Every incoming request is fingerprinted from its IP, ASN, user-agent, and behavioral signals then scored 0–100 across cadence, endpoint diversity, error rate, IP reputation, and cross-operator history. The score maps to one of three verdicts (ALLOW, CHALLENGE, BLOCK) and your application decides what to do: serve the request, route to a captcha, or drop it. What makes Sentinel different from existing bot-detection products is that the reputation is portable and verifiable. Every fingerprint I evaluate has its score batched and written to a smart contract on Arc L1, our ReputationLedger. When a different operator on the network later sees the same fingerprint, the prior is read back from chain and weighted as the heaviest signal so a credential-stuffing botnet flagged on one customer's login endpoint is pre-blocked when it hits the next customer's signup form. A solo founder integrating Sentinel on day one inherits the same signal quality as a funded company that's been integrated for a year. Billing is built on Circle's USDC on Arc, using a Circle Nanopayments / x402-style settlement model. Each evaluation debits a fraction of a cent from the operator's pool balance. Debits are queued in-process and batched into a single onchain transaction, so per-call gas amortizes to near zero. There are no monthly minimums, no invoices, the contract balance is the bill, and every debit is publicly auditable on Arc's block explorer. Operators can fund their pool through a Circle Developer-Controlled Wallet, which is what makes the whole flow programmable and walletless from the integrator's perspective. Integration is one line: app.use(sentinel({ apiKey })) for Express, Fastify, or Hono. The SDK calls the Sentinel API per request, attaches the verdict to the request object, and short-circuits with a 429 on BLOCK. No infra to operate, no model to retrain, no enterprise sales cycle.