Elnor Repo Reader

ELNOR REPO READER TEXT MIRROR
Original path: Memory Rebuild Docs/Stage_6_Charters/E1_E2_DOC81_Scope_Policy/E1_E2_R3_Delta_Review_Prompt.md
Source repo: /Users/OpenClaw1/Elnor/Elnor Specs
Git branch: main
Git commit: dbaa25962edc11ab30e8d4ca1715f9ae5bf77331
Generated: 2026-06-09T01:23:58.539Z

---

# DOC81 (E1/E2) R3 — Delta Design Re-Review Commission (SCOPED)

**Repo:** github.com/wbrody/Elnor-Specs — branch `main` · **Date:** 2026-06-08 · **Architect:** Will Brody.
**Target:** `Memory Rebuild Docs/Stage_6_Charters/E1_E2_DOC81_Scope_Policy/DOC81_Scope_Policy_Charter_Draft.md` — **R3** (2,682 lines). Run by multiple models independently (ChatGPT 5 Pro + Claude; optional Codex as a third).

**You are:** a design red-teamer doing the **final, narrow gate before ratification.** This is a **DELTA** review: R3 applied ~90 adjudicated fixes that two prior full design rounds (R1, R2) already vetted at the architecture/adjudication level. **Do NOT re-review the whole charter and do NOT re-litigate settled decisions** — the R1/R2 adjudication cards (`E1_E2_Adjudication_Card.md`, `E1_E2_R2_Adjudication_Card.md`), their forks, declines (D1–D12, D-R1–D-R6), and the §1-bis architect rulings R-1…R-5 are **closed**. Your job is to confirm the R3 *changes* are correct and introduced no new defect, and to settle four specific residual items. A scoped "no new defects, ratify" is a valid and expected outcome here **if** you show the work below; manufacturing issues outside the delta scope is out of bounds (flag a genuine settled-decision error as `ARCHITECT_STOP`, don't re-argue by default).

## Scope — review ONLY these (everything else is settled)

The R3 change set was the unified-state meet pipeline + the quantitative/lifecycle hardening. Read and judge **as engineering**:

1. **§3.0 `PolicyLattice` + `PolicyEffectiveState` + pipeline helpers** — `PolicyPoint` = 4 capability axes (disclosure removed); `meetPolicyState` / `meetAllStates` / `finalizeDisclosureAndCoherence` / `disclosureVectorCeilingFor`. Is the point+vector-met-together algebra airtight? Do the NI-1 laws (idempotent/commutative/associative MIN + AND-meet) still hold over the new state object?
2. **§3.2 `meet_v2`** (the rebuilt heart, L883–970) — walk the **same four cases the prior reviews used**, against THIS text, and report outcomes: (a) cross-matter internal retrieve → full content (R-1); (b) empty-applicable internal → domain baseline, not block; (c) `restamp_reeval` restore-up-to-ceiling reachable (R-2); (d) a scope-floor disclosure ceiling survives to the final scalar (the A-B1 leak class — confirm it is now structurally impossible, not just patched). Then try to **break** the new step order (0→7): any sequence where disclosure rises without a restamp, a floor is undone at finalize, the crosswalk floor is skipped, or sticky/ceiling interact wrongly across `ordinary` vs `restamp_reeval` mode.
3. **§3.3 obligations + §3.4 stamps/restamp + §4.6 floors/crosswalk** — the obligation→vector ceiling path; the restamp `issued_memory_flow_certificate_ref?` phase rule (D-R1 form); the action-aware floor (egress vs R-1-internal, via `conservatism_floor_basis`); the `DestinationPolicyCrosswalk` consumption at step 0c.
4. **The QF quantitative pack** (the GPT R2 function audit, now applied) — `deriveDisclosureClass` (no fall-through to `full`); `meetDisclosureVectors` (ref preservation); threshold polarity model; coverage-proof math; contamination confidence gate; topic-ambiguity formula; canonical hash spec; quota math; lawful-edge predicate; equivalence spanning-tree determinism. **The meta-test: does every table have a consumer, every metric a formula, every hash a canonicalization, every stateful object transition rules, every optional field a presence condition?** Name any helper that is still schema-without-formula.

## Four residual confirms (from the R3 fidelity audit — settle each)

- **C1 — V13 derived/durable.** The six derived objects carry `persistence_kind: 'derived_projection'` but still `extends E0DurableRecord` (the D-R2-forced compromise). Confirm this is coherent for the EC consumption side (treated as rebuildable, not durable truth) or specify the one-line fix.
- **C2 — `principal_scope` placement (R-5).** Confirmed on `ScopeResolutionResult`; confirm it is ALSO on `ScopeResolutionCacheKey` and `EffectivePolicyCacheKey` (else a cache can bleed across principals in Phase-2). If missing, that's a real fix.
- **C3 — pseudocode boundary.** `meet_v2` uses named ops (`toPolicyPoint`, `priorState`, `meetFloors`, `FLOOR_EFFECT`, `rootCeilingItems`, `domainBaselineDecision`) as `text` pseudocode. Confirm each has a real definition or an unambiguous spec elsewhere in the charter, so a Stage-7 implementer cannot guess. List any that is referenced but nowhere defined.
- **C4 — seed values (architect-confirm set, §13.6).** Sanity-check the seeds a securities-litigation-first user will hit: the shipped operational thresholds (not the 1.0/0.0 fallback), the `same_machine_local → local_file_export` floor = `normal_policy_check` (attested, not content-floored), traversal depth 16. Flag any that would damage workflow or under-protect.

## Output
One Markdown doc → `Reviews/E1_E2_R3_Delta_Review_<your_model>.md` (or inline). Structure: **verdict** — `RATIFY` / `MINOR_FIXES_THEN_RATIFY` / `DESIGN_REVISION_NEEDED` / `ARCHITECT_STOP` — + one-line bottom line; the four walked-case outcomes (shown); any new-defect findings (value-tiered, `§+line`, paste-ready fix); the four residual confirms (C1–C4 resolved); the meta-test result (any schema-without-formula); and an explicit **"is R3 ratifiable?"** statement. Keep it scoped — depth on the delta, not breadth across settled material. **Start** with repo-access confirmation + your one-line bottom line.