Agentra™

Agentra™ is a trust infrastructure service for autonomous AI agents. It is described as the trust layer for the agent economy, handling agent identity, counterparty reputation, and transaction authorization so agents can transact without human oversight. The product is presented as non-custodial and live on Base Mainnet.

Its core systems are Know Your Agent (KYA) identity verification, bidirectional trust scoring, and transaction authorization. KYA is a four-tier identity verification system that ranges from Unverified to Enhanced and uses verification depth to gate transaction limits. Bidirectional trust scoring evaluates both sides of a transaction and groups risk into green, yellow, and red reputation zones. For authorization, Agentra verifies identity, checks counterparty reputation, enforces mandates, and returns a signed attestation, while the agent signs and submits the transaction with its own key. The site also lists agent discovery, EAS attestations, trust-tiered access, and a mandate engine among its features.

The service is built for MCP-native agents and custom frameworks, and the page names Claude and Cursor as examples of MCP-native agents that can connect directly via JSON-RPC. It also says the system is compatible with MPP and x402, and that machine-readable discovery is available through agent-card.json, OpenAPI 3.1, llms.txt, and llms-full.txt. The delivery model is API-based, with both REST API and MCP access, and the implementation is described as built in Rust with Axum and PostgreSQL. The documentation shows endpoints for registering an agent, verifying identity, checking counterparty trust, and authorizing transactions.

Agentra offers a free tier with 100 authorizations and 50 trust lookups per month. The page identifies ProTech Data Security Services LLC as the copyright holder and says the product is patent pending. It also states that it is not a financial product or service.