A vocabulary anchor for judgment-aware agents — systems that distinguish between current judgment and stale opinion in their persistent memory, and refuse to silently overwrite stable user judgment via auto-mutation.
The opposite of memory-as-bag-of-text: memory as a structured substrate where judgment integrity is a first-class invariant.
This name refers to a position outlined in detail at:
- coevolution-kernel — main thesis
- claude-cognition-skills — reference implementation
This repository is a vocabulary anchor; the substantive content is in the linked thesis and implementation.
License: MIT.
Code and documentation in this repository are licensed under the GNU AGPL-3.0 (see LICENSE) as of 2026-09-13. Earlier releases were published under MIT and remain available under those terms. The names "Tuppo", "途朴", "Ceruvon", and "naze" are not licensed — see NOTICE.