Basilisk Labs — Experimental AI infrastructure and cognitive architecture studio.

We build local-first systems for autonomous agents whose work, memory, beliefs, and decisions must remain inspectable over time.

Current work includes AgentPlane, a Git-native audit layer for coding agents; DEUS on OpenClaw, a cognitive runtime layer for explicit memory, beliefs, introspection, and action gating; and cognitive graph research for connected agent state-transition artifacts.

Agent context: llms.txt

GitHub: github.com/basilisk-labs

Mail us: hello@basilisk-labs.com