—

Implementations

aindf — verify
$ aindf verify ./malevich/conformance.json
malevich
reference implementation · awaiting publish
target 6/6
checking 6 principles…
Principles
01semantic-namingtarget 02machine-contractstarget 03closed-applicabilitytarget 04slot-contractstarget 05single-sourcetarget 06agent-surfacetarget
→ target: 6 / 6 · fully ai-native (on publish)
$ submit an implementation ↗

Conformance is a score, not a binary. A system declares which principles it satisfies; the validator confirms or refutes against spec §8 and emits a gap report. Partial counts — and comes with a roadmap to the rest.

  1. [ref]
    malevich · reference implementation soon

    Built alongside the spec, co-evolving with it. Targets 6 / 6: semantic naming, machine-readable contracts, closed applicability, slot types, single-source build, and an MCP surface with validated conformance. Published once it has proven the specification.

    ↗
  2. [+]
    your DS here · declare coverage

    Point the validator at your declared conformance. Open a PR. The score is whatever the validator returns — partial counts.

    ↗
  3. [+]
    your DS here · 3 / 6 still earns a badge

    Declaring three of six principles earns a badge and a roadmap to the missing three. The validator outputs gap reports, not just a yes/no.

    →
  4. [+]
    your DS here · conformance is versioned

    A 6 / 6 against v0.1 is recorded against v0.1; v0.2 may add a seventh principle. The dependency arrow points one way: implementation → design system → framework.

    →

A design system is fully ai-native when it ships all six principles from one source. Most ship a subset. aindf is the spec for the whole graph.

$ git clone …/ai-native-design-framework.git