An open protocol for verified human identity, delegated agent authority, and full on-chain auditability. Built on Solana.
As AI agents become more autonomous, the question shifts from "what can they do?" to "who authorized them and can we verify it?" HumanRail provides the missing trust layer — a protocol where every agent traces back to a verified human, every action is bounded by explicit capabilities, and every transaction leaves an immutable receipt.
Every agent is anchored to a verified human identity. No anonymous bots. Full accountability chain.
Capabilities define exactly what an agent can do — spending limits, program scopes, time windows, and risk tiers.
Every action produces an on-chain receipt. Compliance-ready audit trails from day one.
Four layers. Seven programs. One unified trust stack.
Verified human profiles with attestation-based trust scoring and third-party identity proofs.
AI agent profiles linked to human principals with key rotation, TEE verification, and lifecycle management.
Capability tokens with per-transaction limits, daily caps, program scopes, risk tiers, and emergency freeze.
PDA-controlled escrow payments with invoice creation, capability-checked authorization, and dispute resolution.
On-chain document signing with hash-based verification and multi-signer attestation support.
Immutable action logs for every protocol interaction. Full audit trail for compliance and analytics.
From identity primitives to a full ecosystem of composable modules.
Agent spending with escrow and limits
Request and fulfill agent services
Multi-party on-chain attestation
TEE and signed build verification
HumanRail is live on Solana devnet. Connect your wallet, create your identity, and register your first agent.