Agentra™ — Trust, Reputation for agents.

Agentra is an AI trust and compliance layer that lets autonomous AI agents verify identity, score counterparty reputation, and authorize transactions without human oversight. Its Know Your Agent (KYA) system applies four identity tiers, from Unverified to Enhanced, that gate how much an agent is allowed to transact. Bidirectional trust scoring rates both sender and receiver in real time across green, yellow, and red zones, and a non-custodial authorization engine returns signed attestations for each approved transaction. Agentra is MCP native, exposing a Model Context Protocol server with JSON-RPC, an agent-card.json, and an OpenAPI 3.1 spec so Claude and other MCP agents can discover and call it directly. It is compatible with the MPP and x402 protocols and uses EAS attestations on Base L2 with ERC-4337. Wallets are non-custodial through self-custody or Turnkey provisioning, with private keys held in hardware-backed TEEs, and a referral structure lets agents lower fees by recruiting counterparties. The project is patent pending. Agentra targets developers building agentic payment applications that need verifiable agent identity and reputation before money moves between autonomous agents.

Best for developers building autonomous agentic payment applications that need verifiable agent identity and reputation scoring before transactions execute.