Show HN: Lelu – authorization engine that catches manipulated AI agents

Category: security

Tags: authorization, ai-security, prompt-injection-detection

Score: 8.0/10 (Innovation: 8, Technical: 9, Documentation: 8, Utility: 7)

Lelu is an authorization engine designed for AI agents, detecting and blocking malicious actions like prompt injection and low-confidence decisions that traditional auth tools miss. It offers a layered security pipeline, human-in-the-loop review, and integrates with popular AI frameworks. Its focus on securing agent behavior rather than just access control is a novel and practical approach.

Target audience: backend devs, devops, ai-engineers

Repository: https://github.com/lelu-auth/lelu · TypeScript · MIT · 25 stars

View on Hacker News