lictorate
The enforcement layer for AI agents.
We build the runtime infrastructure — policy as code, audit at every call — that keeps powerful agents within the bounds you set.
thesis
AI agents are gaining capabilities the security model wasn't built for.
Tool calls. Browser control. Code execution. Autonomous decisions over real systems with real consequences. We build the layer that decides what agents can do — at runtime, in code, with policy you can read.
principles
Principles
Authority before action.
Policy precedes execution. No model output reaches a tool until the rules permit it.
Defense in depth.
One prompt injection should not be catastrophic. Enforcement is the layer that contains the blast radius.
Auditable enforcement.
Every decision visible. Every rule reviewable. Security teams should read code, not hope.
notes — first essays shipping soon
Notes
- Apr 2026Why enforcement, not detection
- Mar 2026AgentGuard v0.4 ships MCP adapter
- Feb 2026On policy as code
what we ship
What we ship
AgentGuard. An open-source runtime firewall for AI agents.
Apache 2.0. Core in Go. Adapters for LangChain, CrewAI, browser-use, and MCP.
built by