CogitaveLearn
Training

Summary

You can now explain why Cogitave is building its own sovereign kernel from scratch, and how to judge a verification claim instead of taking it on faith.

In this module, you:

  • Compared the two layers directly: Namzu is TypeScript, runs on Node, and you use it today; Yuva is no_std Rust, boots on hardware as the machine itself, and is not yet something you host a workload on.
  • Explained why an isolation, memory, or audit guarantee is only as strong as the layer enforcing it, and why Yuva - a from-scratch, no_std Rust kernel and micro-VMM - exists to own the hardware directly rather than sit as a process inside someone else's operating system.
  • Learned the four things that make Yuva's verification claim checkable: unsafe code confined to one audited HAL crate, selected leaves formally proved with Kani (with harness counts pinned in CI, so a shrinking proof set fails the build), a dual-architecture (x86_64 / aarch64) boot exercised on every push, and a boot that fails closed - a missing self-test marker is a failed boot, not a warning.
  • Learned to read Yuva's own honesty tokens: uppercase status markers reported over serial that state plainly which capabilities are not live yet.
  • Stated Yuva's current status accurately: live inference is MOCK and learning is DORMANT (gate not met) - Yuva today is a verified boot and kernel substrate, not a runtime you can deploy an agent onto; Namzu on Node is where you run an agent today.

That completes the two modules of the "Agent platform fundamentals" path: building your first agent with Namzu, and now understanding the sovereign kernel underneath it.

Next steps