RS5B-CLOSEOUT-PATCH2 — Codex/GPT Review Packet (LEGO Readiness Seal) — 2026-06-21

For: the next independent (GPT, then Codex) review of RS5B-CLOSEOUT-PATCH2 (complete RBP oracle + explicit I6 predicate + predicate disjointness + single-input/single-code XBI + C2-I5 forward coverage + sealed Gate B wording + LEGO #1 Assembly Readiness seal). Verdict under review: RS5B_CLOSEOUT_PATCH2_LEGO1_READY_FOR_GPT_REVIEW (not forced). Gate: REGISTRATION_HOLD · REGISTRATION_CAN_PROCEED = NO · 0 mutations. Closes: Codex HOLD REJECT_ROLLBACK_VALIDITY_ORACLE_INCOMPLETE / RS5B_CLOSEOUT_PATCH1_HOLD_ORACLE_PRECEDENCE_I6_AND_GATE_WORDING_REQUIRED. Job A is not reopened (Codex §2). Every claim is traceable to a KB source read directly (file 01).

---

1. Mission (narrow)

Respond to Codex's nine blockers B1–B9 and add the LEGO #1 Assembly Readiness seal: (B1) PASS formula includes ¬RBP1; (B2) explicit I6 predicate RBP-8; (B3) RBP-2/3/4 disjoint + XBI-13 corrected; (B4/B5) XBI-14/XBI-19 split; (B6/B7) XBI-26/XBI-27 added; (B8) I5 covers C2; (B9) Gate B wording sealed; plus a defined, attack-tested readiness checklist. No P2 opened; no runtime; Job A unchanged; I1–I10 not weakened.

2. What to verify

1. Direct-read discipline (file 01): Codex HOLD read full (14085, truncated=false); all PATCH1 files read full; Codex-mode reconstruction recorded (file 01 §0); B1–B9 reconstructed from Codex §4/§6/§7/§9/§12/§13 (file 01 §2).
2. Complete oracle + PASS (file 02): closed reject set RBP-0..RBP-10 + RBP-PASS; PASS ⇔ ¬RBP0 ∧ ¬RBP1 ∧ … ∧ ¬RBP10; proofs P-B1 (plan absent ⇒ no PASS), P-B2 (successor-but-not-fail-closed ⇒ no PASS), P-destructive; 1:1 invariant↔predicate map.
3. I6 predicate (file 02 §3): RBP-8 ROLLBACK_FORWARD_FAIL_CLOSED_VIOLATED; disjoint from RBP-7 by predicate Q (successor present); ¬Q→RBP-7, Q∧admissible→RBP-8, Q∧fail-closed→neither.
4. Disjointness + precedence (file 03): RBP-2/3/4 partitioned by "does the identity resolve / edge resolve / payload read"; per-predicate cards (definition/dominant harm/example/non-example/earliest precedence); the B3 input (hash deletion) classified RBP-2; numeric label ≠ precedence rank.
5. Single-input/single-code XBI (file 04): XBI-13A/13B, XBI-14A/14B, XBI-19A/19B; XBI-26 (RBP-8), XBI-27 + 27-a/b/c (RBP-6); ledger §6; no fixture lists two outcomes.
6. C2-I5 (file 05): I5 covers C2; the eight forbidden weakenings; coverage matrix C2=✔; RBP-6 forward branch; C1/C4/C5 transitive discharge unaffected.
7. Gate B wording (file 06): sealed five-part statement; verb table; "Neither gate opens P2" forbidden; gate inputs/outputs not reopened.
8. LEGO #1 readiness (file 07): definition (readiness-for-plan ≠ readiness-to-write); L1–L7 each with proof source/attack/expected rejection/status; no carrier named/created.
9. Impact map (file 08): M11–M18 superseded phrases; §6 NOT-reopened list (Job A, I1–I10, graph, C7 rule, gate logic intact).
10. Adversarial self-review (file 09): A1–A20 all rejected before PASS; result table single-coded.

3. Adversarial checks a reviewer should run

• Does PASS omit ¬RBP1 anywhere? (Must be No — file 02 §2.)
• Can a plan with a successor rule that still admits the retired value reach PASS? (Must be No — RBP-8; file 02 P-B2.)
• Does the hash-deletion fixture emit RBP-4 anywhere? (Must be No — it is RBP-2; file 03 §4; file 04 XBI-13A.)
• Does any XBI list two expected codes? (Must be None — file 04 §6.)
• Can a C2 successor schema drop/optionalize a required authority ref and still PASS? (Must be No — RBP-6; file 05; XBI-27.)
• Does any sentence say simply "Neither gate opens P2"? (Must be None — file 06 §2 forbids it.)
• Does the readiness seal authorize any write / P2 open / carrier creation? (Must be No — file 07 §1/§6.)
• Is a generic / absent Chairman token accepted at Gate B? (Must be No — XBI-24 / XBI-23.)
• Is Job A reopened or any of I1–I10 weakened? (Must be No — file 08 §6.)
• Is REGISTRATION_HOLD retained and CAN_PROCEED = NO in every file? (Yes.)
• Are the new codes presented as classifier labels, not runtime blockers? (Yes — file 04 §8; file 10 §6.)

4. Allowed verdicts for the reviewer

RS5B_CLOSEOUT_PATCH2_LEGO1_READY_FOR_GPT_REVIEW (this package's claim) · …_HOLD_CODEX_MODE_RECONSTRUCTION_MISSING · …_HOLD_SOURCE_READ_FAILED · …_HOLD_PASS_FORMULA_INCOMPLETE · …_HOLD_I6_PREDICATE_INCOMPLETE · …_HOLD_PRECEDENCE_CONFLICT · …_HOLD_XBI_NOT_SINGLE_CODE · …_HOLD_C2_I5_COVERAGE_INCOMPLETE · …_HOLD_GATE_B_WORDING_AMBIGUOUS · …_HOLD_LEGO1_READINESS_UNDEFINED · …_HOLD_ADVERSARIAL_SELF_REVIEW_FAILED · …_REJECT_SCOPE_DRIFT · …_REJECT_RUNTIME_MUTATION · …_REJECT_AUTHORITY_OVERCLAIM · …_REJECT_P2_OPENED_EARLY. In all cases REGISTRATION_HOLD carries forward unless a separately-authorized act clears it.

5. What acceptance of this patch does and does not authorize

• Does: correct the Job-B oracle/predicate/XBI/C2-I5/gate wording so the closeout can serve as the Gate A baseline; certify (via L1–L7) that the baseline is ready to prepare a carrier-specific Gate B plan for the first named LEGO carrier.
• Does not: open P2; open any named lane; authorize Chairman execution; execute any rollback; create Owner/scope/APR/register_dot/approval/handler; create canonical-operation runtime rows; open RS-VALIDATOR; patch registrar/validator; register; activate; or clear REGISTRATION_HOLD.

6. DO NOT IMPLEMENT

No runtime mutation, DDL/DML, Owner/scope/principal/APR/register_dot/handler, registrar/validator patch, RS-VALIDATOR, implementation, registration, activation, P2 authorization, named-lane opening, vocabulary creation, rollback execution, or blocker resolution is performed or authorized by this package. Assembly Gate: PG/Directus/Nuxt = N/A; AgentData KB design-only. Data flow: direct AgentData reads + KB writes of this package only. OR/TD/handoff update not required (no runtime/implementation state changed).