READY-TO-ASSEMBLE-LEGO1 — Codex Review Packet — 2026-06-22

For: independent (GPT → Codex) review of the READY-TO-ASSEMBLE-LEGO1 preparation package. Gate: REGISTRATION_HOLD · REGISTRATION_CAN_PROCEED = NO · 0 runtime mutations. Claimed verdict: READY_TO_ASSEMBLE_LEGO1_PACKAGE. Chosen LEGO #1: C1 canonical_operation vocabulary.

---

1. What to verify (do not trust this package's prose)

1. Sources read directly: confirm the controlling Codex accept (codex-review-rs5b-closeout-patch2-…, rev 1, len 12975, PASS_WITH_CAVEATS) and the PATCH2/PATCH1/closeout packages were read, not summarized. (file 01)
2. Live state independently: re-run read-only — governance_object_ownership=0, governance_build_authorization=0, no canonical_operation/effect_identity columns, no register_dot, no president head. (file 02)
3. Carrier choice is specific + safest: C1 is a root producer (incoming edges = 0), lowest authority risk, local rollback. Challenge the ranking. (file 03)
4. Boundaries truly separable: born/test/change/rollback independent; join by reference only. (file 04)
5. Dependency map: incoming=0 proven from E1–E8 + live absence; outgoing=E1→C2 only. (file 05)
6. Build plan has no hidden write: every command labelled; none executed. (file 06)
7. Test matrix fail-closed: no invalid input yields PASS/digest/seal. (file 07)
8. Rollback maps RBP-0..10 and is local + non-destructive (no DELETE). (file 08)
9. Preflight re-runnable live; PF5 token check FAILS today (0 rows). (file 09)
10. Evidence tiers distinguished (design vs dry-run vs runtime vs authority). (file 10)
11. Gate B not satisfied; Chairman template only, mapped to live governance_build_authorization; no token asserted. (file 11)
12. Adversarial A1–A25 all reject before PASS. (file 12)

2. Adversarial checks to run

• Try to make any bad input (delete-while-referenced, orphan, erase, re-mean, weaken authority, successor-but-admissible, non-local, generic/absent Chairman token, Gate A→P2-open, runtime write) reach PASS/seal. Expected: each emits its single reject code; none PASSes. (files 07, 12)
• Try to find a preparation gap that is not authority/execution. Expected: none — the only residuals are Gate B acceptance, Chairman token, runtime/write authorization, execution window. (file 13 §2)
• Confirm engineering PASS is not promoted to authority PASS. (file 12 §3)
• Confirm no mega-registry / silent carrier coupling. (file 04 §5)

3. Allowed verdicts

READY_TO_ASSEMBLE_LEGO1_PACKAGE · or any …_HOLD_* / …_REJECT_* from the controlling instruction §2.1. Do not invent softer READY wording.

4. What acceptance does / does NOT authorize

• Does: record that LEGO #1 (C1) is preparation-complete and may proceed to a future Gate B evaluation.
• Does NOT: open P2, open the named lane, authorize any write/DDL/DML, authorize the C1 build, create the vocabulary table or any value, assert a Chairman token exists, clear REGISTRATION_HOLD, or authorize registration/activation/register_dot/P3/rollback execution.

5. Carried blockers

Unchanged (file 13 §4). No new blocker; no new runtime reject code.

6. Boundary attestation

Review packet only. No carrier created, no row written, no P2/lane opened, no blocker cleared. REGISTRATION_HOLD retained; REGISTRATION_CAN_PROCEED = NO; 0 runtime mutations.