KCC 14

What KCC Is Not

Explicit scope boundaries to prevent misinterpretation — what KCC structures, and what it deliberately leaves to other tools and to human judgement.

BoundariesScopeBoundariesStructuralEngineering judgment
Created 2026-06-08 · v0.4.0

Explicit Scope Boundaries

Explicit scope boundaries to prevent misinterpretation.

KCC is NOT...Because
a runtimeIt does not execute agents or orchestrate; the kernel is contract specification plus enforcement infrastructure. Cells execute their own invocations.
a model registryIt governs agents, not the underlying models. Model choice is a cell-level decision constrained only by the cost envelope.
an evaluation frameworkIt produces traces and computes confidence, but does not specify how to evaluate model quality. See 10.6.
a prompt management systemPrompts live within capabilities; the kernel does not specify how they are written or stored.
a tool registryCells use whatever tools their agents declare; the kernel validates declarations but maintains no master registry.
a methodologyKCC is structural; it is compatible with Scrum, Kanban, Shape Up, or none.
a security frameworkIt addresses specific concerns but is not a comprehensive security model.
a vendor lock-inThe specification is open; no specific tooling, vendor, or platform is required.
a maturity assessment toolIt is the target architecture, not a way to score where you currently are.
a consulting frameworkIt is published as a body of work; anyone can adopt it without paying anyone.

KCC Is Not One-Size-Fits-All

KCC is designed for organizations with at least three engineering teams adopting AI at scale. A single team can adopt parts of it, but the full model assumes multi-team composition.

KCC Is Not a Replacement for Engineering Judgment

KCC structures decisions; it does not make them. Capability maintainers, kernel maintainers, cell owners, and engineers retain full judgment over their domains. The model provides a vocabulary, a contract, a maturity ladder, meta-agents, a learning pipeline, and a measurement framework. It does not provide the right answer to specific design questions, substitute judgment, replace skilled engineering leadership, or guarantee successful AI adoption.

Frameworks are tools. They shape decisions. They do not make them.