Debugging and retained non-transfer boundaries

• Dynamics precedes output states. All source controls, pulse areas, emission matrices, initial states, and radiation factors are specified before the channel calculation. The off-diagonal matrix unit is evaluated explicitly.

• Degenerate input energy is not thermal selection. Messages have the same initial energy. Energy conservation does not determine $p,\eta$, or the occupation amplitudes.

• Mixed key purification is retained. The key is random only after tracing its independent purifying system. That system belongs to every relevant complementary output.

• The dephasing environment is not silently discarded. Its distinguishable states quantify successful-sector leakage. An observer who lacks it cannot invert dephasing on an unknown entangled input.

• Private marginals are not a private collective channel. Each single cell is a replacement channel. The two-cell channel contains the key–ciphertext correlations needed for recovery.

• Calibration is not Hawking-law derivation. The occupation weights are taken from the specified Em03-A–C baseline and inserted into finite Hamiltonians. The result derives their finite model output, not the original greybody-weighted Bose law.

• Payload degrees of freedom are not hidden in scalar modes. They have their own Hilbert spaces and energy ledger. Scalar occupation matching does not identify the full cell state with an Em03-C two-packet state.

• Diamond recovery and a Bell test are distinct criteria. They have different definitions. Their simultaneous optimal values here follow from a model-specific Pauli and Bell argument, not a silent fidelity-to-diamond substitution.

• Complementary leakage uses a named metric and center. Equation (eq:leakage) includes every inaccessible system. It is not an unrestricted optimization over replacement centers, nor a general equality between leakage and recovery.

• Coherent return precedes destructive verification. Classical records cannot replace the returned quantum message. Metadata can be discarded after coherent correction without measuring the logical output.

• Finite access is newly constructed. The access proof concerns these complete finite cells. It does not extend a two-packet scalar capture theorem to an arbitrary evaporation history.

• Entropy is not a decoder certificate. The same pure-input radiation entropy can accompany different reference correlations and recovery errors.

• Perturbation estimates require complete interfaces. Initial purification, independent preparation, invalid words, failure outcomes, cross-register coupling, and references are part of the stated comparisons.

• The model is not a gravity-generated-entanglement test. The specified non-gravitational finite controls can transfer pre-existing reference entanglement. They do not demonstrate a gravitational mediator or exclude classical gravity models.