RS-TKT-1-PATCH1B · 08 — Supersession and Non-Regression Map

Lane: RS-TKT-1 — Phase 1 TKT Base Design Package · PATCH1B (dry-run readiness preflight / proof-doc-only) Date: 2026-06-22 Gate: REGISTRATION_HOLD · REGISTRATION_CAN_PROCEED = NO · 0 runtime mutations Authority: NON_AUTHORITY · may_gate=false · decision_effect=NONE

PATCH1B is additive. It edits no prior file; every prior file stays at its current revision. This map states exactly what PATCH1B supersedes-by-addition, what it leaves untouched, and what stays closed.

PATCH1B supersedes ONLY proof/readiness defects.
PATCH1B does NOT weaken P1 / P6 / P7.
PATCH1B does NOT reopen P2–P5.
PATCH1B does NOT authorize Phase 2.
PATCH1B does NOT execute dry-run.

---

1. Old Phase-1 claims superseded-by-addition (files stay revision 1; not edited)

Phase-1 location	premature claim	superseded by	nature
17 §5	"negative-fixture matrix COMPLETE"	patch1b/02 + 03 (coverage proven by count: 14/14, 46 fixtures)	proof completion
17 §1	SAFE_REJECT in the status column (BAD-FC-003/005)	patch1b/01 §2 + 02 §1 (status≠outcome; PASS/SAFE_REJECT separated)	schema correction
17 §2	BAD-L3-001 = FAIL / design REJECT (dual code, two layers)	patch1b/02 §7 (BAD-L3-001 design + BAD-L3-002 runtime, one code each)	split
17 §3	BAD-NVSZ-002 = ESCROW_E9 / ROOT_E4 (two namespaces)	patch1b/02 §4/§7 (ESCROW_E9 only; ROOT_E4→BAD-NVSZ-003 Phase 3)	split
17 §4	BAD-RS-001/BAD-PROP-001/002 prose/config-error results	patch1b/01 §3 + 02 §7 (one DESIGN_LINT/RS_* code each)	code assignment
16 §3	"compatibility matrix COMPLETE / deterministic"	patch1b/04 (per-row canonical fixture + per-row effects + dryrun_relevance)	proof completion
18 §4 / 19 §5	design/blueprint "PASS" asserted by citation	patch1b/06 + 07 (count-based + active self-break)	method upgrade
22 §7/§8/§9/§14	"complete / single result / no reviewer interpretation"	patch1b/01–07 (the canonical machine-counted proof)	proof completion

These were already retracted-by-supersession in patch1/00 §3. PATCH1B re-states the full set and adds the dry-run-readiness proofs.
All of 16/17/18/19/22 remain revision 1; the authoritative fixture/oracle + readiness proof is now patch1b/01–09 (inheriting patch1/01–07).

2. PATCH1 claims superseded-by-strengthening (PATCH1 files stay revision 1; reused, not retracted)

PATCH1 location	PATCH1 content	PATCH1B strengthening
patch1/01	canonical schema + catalog (status≠outcome; one code/namespace/layer)	split into master schema 01 + master catalog 02; added fixture_type disjoint partition (01 §1.2), stop_state_if_failed, code-uniqueness invariant (01 §3.3)
patch1/02 + patch1/03	coverage ledger (14 bricks) + 280 contract fields (separate files)	merged into one ledger 03 + added the five LEGO boundary present-checks (70/70)
patch1/04	regenerated traceability + per-row effects	added per-row dryrun_relevance column (04)
patch1/05	DR-1..DR-17 classified; "only Owner authorization remains"	promoted to a full launch-readiness packet 05 (named dry-run, safe-default paths/permissions, stop states, launch gates, pseudo command sheet)
patch1/06	20 zero-defect counts + adversarial probes	expanded to 23 machine counts (07) + a dedicated active self-break file (06) with 5 found-and-fixed ambiguities

PATCH1's verdict RS_TKT_1_PATCH1_READY_FOR_GPT_CODEX_REVIEW is not retracted; PATCH1B raises the same proof to the new
"Codex-confirms-not-discovers" standard. Nothing in PATCH1 is weakened.

3. Accepted PATCH2 rules — UNCHANGED (no weakening)

P1 (PATCH2 01): six-conjunct invalid_input_safe; reserved-token floor; safe-rejection context; full scan surface; detector order;
                deterministic vocabulary (SAFE_REJECT / FAIL_* / HOLD_OUTPUT_SURFACE_UNAVAILABLE). PRESERVED verbatim in 02 §1 + 06 §2.
P6 (PATCH2 02): profile_id ∈ {CODEX_CAUGHT_RS5A, SELF_REPORTED_RS5B_DRAFT}; scope_class ∈ {STRUCTURAL_COMMON, RS5A_SPECIFIC, RS5B_SPECIFIC};
                no profile_id=structural; no automatic promotion. PRESERVED in 02 §5 + BAD-RS-001.
P7 (PATCH2 03): five status fields; L0→L1→L2→L3 dependency; FAIL & HOLD propagation to N/A; FAIL>HOLD>PASS; 10-row truth table;
                aggregate advisory; effects always NONE. PRESERVED in 06 §7-aligned distinction checks + BAD-PROP-001/002.
⇒ P1/P6/P7 weakening_findings = 0.

4. Codex-confirmed P2–P5 / boundaries — NOT reopened

P2 four-way L3 split (AUTHORITY-FIREWALL / CLAIM-AUDIT / IDENTITY / NVSZ) ... NOT reopened; 4 independent bricks in 03.
P3 source-authority Tier-1/2/3 hierarchy ................................... NOT reopened; carried as MCB-6 caveat (04 row).
P4 L1 inert-only / HOLD_RUNTIME_SURFACE_REQUIRED ........................... NOT reopened; BAD-L1-001 preserved.
P5 NVSZ ESCROW_E*/ROOT_E* taxonomy + canonical hash_manifest.sha256 ........ NOT reopened; BAD-NVSZ-001/002/003 preserved.
Codex §11 P1 PASS · §12 LEGO PASS · §13 authority PASS ..................... confirmed, not reopened (00 §2).

5. Caveats — remain non-blocking (Codex §14)

MCB-1 RS5B no external Codex review ⇒ SELF_REPORTED_RS5B_DRAFT ... non-blocking for Phase-1 / Phase-2-readiness; DEFERRED (04).
MCB-5 NON_VECTOR_ROOT undesignated ............................. blocks Phase 3 only; DEFERRED (BAD-NVSZ-003, ROOT_E4).
MCB-6 no enacted laws-new architecture baseline ................ non-blocking; Tier-1>Tier-2>Tier-3 meanwhile; OPTIONAL (04).
"0 runtime mutations" = package attestation, not live PG/Directus proof ... unchanged.

6. Phase-2 gates — remain CLOSED

Phase 2 is NOT opened by PATCH1B. The two remaining gates are:
  GATE-1 Codex confirmation of PATCH1B            (CODEX_CONFIRMATION_REQUIRED)
  GATE-2 Owner/GPT explicit "open Phase 2" command (OWNER_AUTHORIZATION_STILL_REQUIRED)
Invariants that must stay true: REGISTRATION_HOLD active · REGISTRATION_CAN_PROCEED = NO · authority_effect/registration_effect = NONE.
No implementation, dry-run execution, runtime, registration movement, or production is authorized.

7. Net statement

PATCH1B supersedes ONLY proof/readiness defects (Phase-1 16/17/18/19/22 premature claims; PATCH1 proof granularity).
It does NOT weaken P1/P6/P7, does NOT reopen P2–P5, does NOT authorize Phase 2, and does NOT execute a dry-run.
weakening_findings = 0 · reopened_closed_items = 0 · prior_files_edited = 0 (all remain at current revision).