HN Showcase
Back to browse
GitHub Repository

Cryptographic proof of human authorization for AI agent actions

5 starsRust

AgentMint – Cryptographic proof of human approval for AI agent actions

by asmhacker·Feb 27, 2026·1 point·1 comment
Visit ProjectView on HN

AI Analysis

●●●BangerBig BrainZero to OneSolve My Problem

Cryptographic proof of human approval for agent actions—solves a real gap in agent safety architecture.

Strengths
  • •Single-use tokens with JTI tracking prevent replay attacks at the cryptographic level, not application logic
  • •Ed25519 signatures verify in 3ms, enabling high-throughput agent systems without auth bottlenecks
  • •Audit trail covers sub, action, timestamp, and jti—exactly what compliance auditors need for agentic governance
Weaknesses
  • •Early stage (v0.1.0) with minimal integration examples; unclear how to wire into existing agent frameworks
  • •No integration patterns shown for Anthropic API, OpenAI, or other agent runtimes—users must implement the bridge
Category
Security
Target Audience

AI engineers building agentic systems with high-stakes actions, financial services, healthcare compliance teams

Similar To

Signers (cryptographic auth) · OAuth 2.0 authorization codes (but shorter-lived and action-specific)

Post Description

Session auth says "agent can use Stripe." It doesn't say "Alice approved this $50 refund at 10:03am." AgentMint signs short-lived tokens proving a human approved a specific action.

Ed25519 signatures Single-use (JTI tracking) 60s TTL default 3ms verify Full audit trail

Built in Rust. Drop-in sidecar for your backend. https://github.com/aniketh-maddipati/agentmint

Similar Projects