Boilerplate agent trait, types, and llmos CLI entrypoint.
Safety-critical cognitive safety library for AI agents. 4-tier architecture (Resource Body, Kernel, Working Memory, Sifter) with formal verification primitives, detection layer, and integration primitives.