Aegis — AI Agent Governance

Aegis is an open source governance layer for AI agents, providing a unified runtime to enforce policy, guardrails, trust, and audit across 12+ major agent frameworks. It auto-instruments existing AI agent frameworks (like LangChain, CrewAI, OpenAI, Anthropic, LlamaIndex, DSPy, and MCP) to add deterministic guardrails for prompt injection, PII masking, trust delegation, policy enforcement, audit trails, and more, all without requiring changes to agent code. Aegis is ideal for teams and organizations building or deploying complex multi-agent AI systems who require governance, compliance, and security controls across diverse frameworks.