Show HN: AgentTrust – Zero-trust security and human approvals for AI agents

Category: security

Tags: ai-agents, zero-trust, security, human-in-the-loop, audit-log, post-quantum

Score: 6.0/10 (Innovation: 6, Technical: 7, Documentation: 5, Utility: 6)

AgentTrust is a zero-trust security control plane for autonomous AI agents, intercepting tool calls to enforce fine-grained policies, scan for prompt injections, gate high-risk actions behind human approvals, and produce a cryptographically verifiable audit log with post-quantum signatures. It combines several emerging ideas—agent capability scoping, blast-radius scoring, and hardware-sealed audit trails—into a single gateway, which is a compelling direction as agent deployments grow.

Target audience: backend devs, devops, security engineers, AI platform teams

Repository: https://github.com/SRN0129/Agenttrust · TypeScript · NOASSERTION · 17 stars

View on Hacker News