HLV-TB-F2P-001 A-GATE-COMP v0.1.0: Pre-Execution Protocol Freeze for Adversarial Apparatus and Provenance Validation

A claim of no-backwards-in-time-signalling (NBTS) violation cannot be supported by statistical significance alone. It additionally requires a physically credible assignment law, auditable chronology and provenance, explicit control of sequential interference and trial selection, a justified joint law across randomization units, and independently certified numerical tail probabilities. This record freezes HLV-TB-F2P-001 A-GATE-COMP v0.1.0, a pre-execution computational adversarial protocol intended to test those requirements before any physical A-GATE or prospective NBTS experiment is attempted. The protocol preserves the historical scope of HLV-TB-F2P-001 V-GATE v0.9.1 and does not retroactively extend its PASS result. It prospectively incorporates the V7-V11 threat families and an extended R12-R28 adversarial fault library derived from independent reviewer-side red-team work. The frozen threat architecture covers shared temporal assignment/outcome structure, forward setting carryover and delayed apparatus memory, latent randomizer/apparatus state, cross-block dependence, post-setting trial retention and missingness, clock and commit failures, entropy buffering or seed reuse, queueing and trial-ID faults, packet loss and retries, stale-cache and partial-write failures, optional stopping, restart selection, multiplicity, custody failures, reset-witness insufficiency, witness common-mode failure, and compound faults. The primary computational design freezes N = 20,000 trials, a primary block size of 250, a two-sided five-sigma threshold of 5.733031437583866 × 10^-7, deterministic seed derivation, valid-null calibration, synthetic positive controls, hard-invalid fault injections, explicit PASS/FAIL/UNRESOLVED semantics, and a three-path numerical certification architecture using FFT only as the fast diagnostic path together with algorithmically independent positive/log-space and arbitrary-precision tail calculations. A candidate anomaly is admissible only if STATISTICAL_PASS, ADMISSIBILITY_PASS, and NUMERICAL_CERT_PASS all hold. Hidden simulation truth may not be used by the decision logic. This deposit contains no A-GATE results and cannot certify a physical apparatus. Hardware-dependent quantities such as reset tolerance, washout horizon, clock uncertainty, commit-latency budget, entropy-source health thresholds, and physical tail-certification tolerances remain explicitly unfrozen for a later hardware-specific protocol identity. A controlled reproduction of the archived V-GATE v0.9.1 author code is a mandatory prerequisite before executing this frozen A-GATE-COMP protocol. Initial pre-execution freeze of the author-run computational A-GATE prequalification. No A-GATE simulation has been executed under this version. The protocol preserves the completed V-GATE v0.9.1 result, incorporates V7-V11 and reviewer-derived R12-R28 threats prospectively, and reserves hardware-dependent certification parameters for a later physical protocol. Any load-bearing amendment requires a new protocol identity.