Skip to content

026-G14 Capability Coverage

Status: PASS_WITH_LIMITATIONS

This audit checks the public capability registry against the controlled qualification and verification records. It covers registry completeness and claim consistency; it does not promote a capability merely because code exists.

Inventory

  • Public capabilities in the registry: 33
  • Public element-analysis mappings: 44
  • Unregistered public capabilities: 0
  • Registry-only capabilities without implementation: 0
  • Duplicate or orphan public capability IDs: 0
  • Unbounded public maturity claims found: 0

No capability is classified as COVERED_FULL. The current public evidence is bounded by gate, element family, formulation, mesh, load, solver route and observable.

Coverage classes

The machine-readable classification is in g14_capability_coverage.json.

Class Count Meaning
COVERED_FULL 0 No unrestricted capability claim is approved.
COVERED_BOUNDED 16 The registry and evidence support a declared bounded scope.
EXPERIMENTAL_ONLY 15 The route is visible and tested or inventoried, but is not a qualified general capability.
NOT_QUALIFIED 2 The route is explicitly outside the qualified scope.
DEFERRED 0 Deferred subscopes are recorded separately rather than hiding a capability.
DEAD_OR_STALE 0 No active capability was removed by this audit.

Important bounded boundaries remain explicit:

  • G07 TL is Owner-qualified only for its tested TET4 domain; HEX8 complete history is not qualified and TET10/HEX20 remain research routes.
  • G08 buckling is first-factor/first-mode and family-specific; HEX8 remains MORE_EVIDENCE_REQUIRED in the active Owner review.
  • G10 finite-kinematic J2 and coupled nonlinear routes remain experimental, deferred or not qualified.
  • G13 external evidence is representative and bounded; non-comparable and superseded records do not support active claims.

Release cleanup

This document records the G14 snapshot and does not itself close the release. At that snapshot, historical full-regression findings, version/test reconciliation, historical documentation metadata, provenance SHA review, stale-route/test inventory and environment-dependent audits were tracked in g14_release_cleanup_items.json. The G15 Owner release review was subsequently completed and is recorded in g15_release_sweep.json.

Full regression was SKIPPED_BY_POLICY for G14; the final release sweep reused the applicable pre-G15 regression evidence.