Fast, deterministic AI tool‑call guard that blocks unsafe actions in <1.5ms
Aegis intercepts every tool call from an AI agent and applies deterministic AST and state checks in under 1.5 ms. It guarantees no network egress, providing cryptographic proof logs and SOC 2/HIPAA‑ready compliance reports. Developers can drop it into LangChain, CrewAI, or any LLM framework with a simple JSON‑RPC hook. Compared to LLM‑based guardrails, Aegis offers instant, provable safety without external calls.
View on GitHub →Snehgabani/aegis-kernel