Lelu – Open-source authorization engine for AI agents

Lelu has launched an open-source authorization engine designed specifically for autonomous AI agents, providing confidence-aware gating, human-in-the-loop review, and policy-driven authorization. The tool allows developers to set confidence thresholds that automatically route low-confidence agent actions to human review, pause agents for approval, and write Rego policies for complex access rules. The engine integrates with major AI frameworks including Vercel AI SDK, LangChain, OpenAI, and Anthropic, aiming to scale from individual projects to production agent swarms.

Lelu. The authorization layer for autonomous AI agents README Authorization that lives inside your agent . Confidence-aware, policy-driven, and built to scale — from weekend AI projects to production agent swarms. npm install lelu-agent-auth npx lelu init Trusted by teams building with Vercel AI SDKLangChainOpenAIAnthropicClaudeMistralLlamaIndexCrewAIAutoGPTVercelFly.ioRailway Features 01 Confidence-aware gating. Every agent action carries a confidence score. Low confidence routes to human review automatically — no code change needed. 02 Human-in-the-loop review. Agents pause, queue their action, and resume once a human approves or denies. Full audit trail included. 03 Policy-driven authorization. Write Rego policies to express complex rules. Allow, deny, or require review — per actor, action, or resource.