Abstract

We develop an audit-hardened finite certificate architecture for a prescribed-host perturbative loop sector, preserving the distinction between mathematical proofs, printed data, supplied scientific records, and execution artifacts. The central v5 finding is a cross-artifact integrity witness: the v4 typeset D01 table has sixth coefficient $-4/3$, whereas its embedded and separately supplied regression has $+4/3$. With the unchanged gauge-table coefficient ${a^{(0)}}_6=4/3$, the regression relation fails by the exact residual $a^{(0)}+f^{\mathrm{reg}}=(8/3)e_6$. Python index 5 is displayed structure $B_6$. Both source values are retained; no physical coefficient is selected by editing a sign.

The supplied v4 transcript reports eight passing suites and one failed suite. Its phase failure occurs before that suite's variation-count and derivative checks, so those later predicates are not described as executed. Their printed-data proofs remain available. A proposed v5 regression checks the exact known discrepancy as a negative control and keeps data-integrity failure separate from successful regression behavior.

The retained mathematical results include complete destructive report instruments with untouched references, the sharp complex qubit-report half-diamond distance ${\max }_j|v_j-{\tilde{v}}_j|/(2B)$, its real binary specialization, original-minor and complete-kernel feasibility certificates, quotient descent versus non-feedback, primitive-enlargement criteria, and an exact affine general-$\xi$ inverse under the declared Q4S kinetic normalization. The old unchecked affine candidate still has residual $(0,0,1{)}^{\mathsf{T}}$; finite-field elimination and even/odd vertex-occurrence controls remain permanent audit components.

The supplied 31-entry second-readout comparison, mutation discrepancies $431/30,1/3,5/12$, 13-of-14 D01 variation, and restricted evanescent rank report $7\to 8$ are preserved with their precise evidence boundaries. The potential impact is a reusable release contract that prevents an expected software failure, a transcription inconsistency, or a missing scientific matrix from becoming a positive physics certificate. No novelty is claimed for elementary channel or rank mathematics. Q03 remains _CLOSED; full gauge rows and counterterm-derivative exactness remain ; the clean-machine rebuild remains . The new appendix is and unexecuted. The cone remains an input, bulk symmetry is not restored, XVI-N remains frozen, and O.1 is _CLOSED outside independently earned scopes.