Conformance Profiles

One schema, three honest bars (roadmap principle P3). A lightweight validator, a governed enterprise deployment, and a live engine should each be able to claim "FLUX" truthfully — at different levels.

Level 1 — Core ✅ definable today

The 0.3.x bar. Any tool can hit it.

• Schema + cross-document validation (the offline reference validator)
• FLUID seam resolves and validates against vendored schemas
• Deterministic under seed
• No extensions required

Level 2 — Enterprise ✅ all requirements shipped as of 0.5.0

Core, plus governance, semantics, and supply chain:

• Extensions declared and typed (RFC-02 — shipped in 0.4.0)
• Lockfile + pinned module refs (RFC-03 — shipped in 0.4.0)
• Playback scorecard present (RFC-05 — shipped in 0.4.1)
• Semantic-model port resolved (RFC-07 — shipped in 0.4.0)
• Signed, attested bundle (RFC-09 — shipped in 0.5.0)

Level 3 — Runtime ◐ contract + vectors shipped in 0.5.0; certification needs a live engine

Enterprise, plus it actually runs and enforces:

• Enforcement contract passes the conformance test vectors (RFC-04 — contract, reference gate and vectors shipped in 0.5.0)
• A live calibration loop feeds the scorecard
• Seam provenance emitted per run (RFC-06 — schema surface + contract-digest check shipped in 0.4.1; per-run emission is engine work)

Why profiles change the calculus

Under a single bar, an advanced implementation is judged against a line it overshoots in some places and misses in others — it reads as "non-conformant" even while doing more than the spec. Under profiles, the same implementation certifies at the standard's top tier, and everything it does becomes a demonstration of the standard rather than a deviation from it. That is how a standard converts its most sophisticated potential adopter into its reference implementation.