# REVIEWED.md — Steward Authorization Record
**Protocol:** David and Claude.app write here after reviewing PENDING items. Claude Code reads this at session start.

---

## REVIEWED-1 — Executor Agency Directive for CLAUDE.md
**Date:** 2026-03-21
**Decision:** AUTHORIZED
**Notes:** Accepted as written with one addition: insert explicit one-sentence 
description of the specific failure mode ("accept self-assessment of criticality") 
immediately after the contamination problem reference. Steward to apply to ~/CLAUDE.md.
This proposal is itself evidence the directive is already operative — the executor 
used the authorization taxonomy correctly on a change affecting its own behavior.

## REVIEWED-2 — Silent Degradation Audit
**Date:** 2026-03-21  
**Decision:** AUTHORIZED — commence after PR merges. Option 1 (full audit, 
concentrated). Produce findings document for steward review before any 
implementation.

## REVIEWED-3 — Factory Schema Idempotency (checkpoint.ts, dedup.ts)
**Date:** 2026-03-21
**Decision:** ACKNOWLEDGED — already in PR scope. The executor's decision to 
extend the fix to the actual error source (checkpoint.ts) rather than the 
originally specified file (job-store.ts) is the correct application of 
"diagnose the class, not the instance." Note this in PR description.

 ## REVIEWED-11 — Approve I15 (ICP-9 Pilot Registry Entry)
  **Date:** 2026-03-23
  **Decision:** AUTHORIZED
  **Notes:** I15 approved as pilot registry entry. l1_contamination_profile field ratified as mandatory for all non-contingent principles. Schema commitment is prospective — all future Cluster A/B entries must include this field. residual_risk field decision deferred pending evidence from pilot operation.your annotation]

  ## REVIEWED-12 — Lodge Design Notes DN-GOV-01 through DN-GOV-04
  **Date:** 2026-03-23
  **Decision:** AUTHORIZED
  **Notes:** DN-GOV-03 requires separate declaration before elevation. DN-GOV-04 is tooling, iterate freely.

## REVIEWED-13 — DN-GOV-05, DN-GOV-06, DN-GOV-07 (Threshold Inquiry Cluster)
**Date:** 2026-03-28
**Decision:** AUTHORIZED with jurist additions
**Notes:**

**DN-GOV-05 (Bounded Self-Repair Principle):** Authorized as DESIGN COMMITMENT. Four-condition framework (reversibility, within parameters, independent verifiability, steward-legible framing) is sound. **Suspension clause:** enforcement mechanics not active until AF-7 reaches design commitment status. The principle is named and load-bearing in constitutional reasoning; it does not authorize autonomous self-repair actions until external review is constitutionally specified.

**DN-GOV-06 (Temporal Authorization Shift):** Authorized as DESIGN COMMITMENT. Mechanism constraints (never reaches amendment, never creates new authority, always conservative direction, no compounding) are correctly specified. **Suspension clause:** same as GOV-05 — not active until AF-7 reaches design commitment status. **Additional constraint (jurist):** temporal shift may not be invoked to justify actions that would not pass the GOV-05 four-condition test in a non-time-pressured context. Time pressure is a trigger, not a relaxation of conditions.

**DN-GOV-07 (The Threshold Already Crossed):** Authorized as FRAMING DOCUMENT. The "threshold already crossed" observation is entered into the constitutional record as an accurate description of current practice. It does not authorize further crossings. Two additions:
1. Working framing ("conditions of trustworthy service") is **explicitly provisional** — may be revised by L1 maturation evidence, Chamber Phase 3+ evidence, or steward philosophical judgment.
2. **Counter-norm on inherited legitimacy:** the fact that a threshold has been crossed does not create precedent for further crossings. Each additional autonomy claim requires fresh warrant, not inherited legitimacy from the prior crossing.

**Cross-cluster consequence:** AF-7 (External Review) is escalated from resolution brief to urgently load-bearing. AF-7 reaching design commitment status is the condition for the GOV-05/06 suspension clause to lift. AF-7 design note should be the next constitutional work item after I17.

## REVIEWED-14 — DN-GOV-08: Constitutional Stabilization, Not Automation of Recognition
**Date:** 2026-03-28
**Decision:** AUTHORIZED — entered as ACTIVE constitutional record
**Notes:** L2 constitutes boundary conditions for recognition; does not automate it. I3 is structural commitment, not capability limitation. Evaluation test added: any invariant touching recognition must be assessed against this commitment before hardening. Phase 3 placeholder (Talmudic chain of transmission, Traditionis) recorded — primary texts not yet in library, blocking condition explicit. Source: Chamber Phase 1, Essay I.

## REVIEWED-15 — I17: Precedence of Present Expression
**Date:** 2026-03-28
**Decision:** AUTHORIZED
**Notes:** Full registry entry following I15 pilot schema. Four-source convergence (Levinas, Fricker, Leopardi, Harrison). Anti-model-disambiguation clause incorporated per adversarial review. Jurist's design question resolved constitutionally: the system does not distinguish genuine divergence from momentary aberration — the interlocutor does. DN-GOV-08 evaluation test passed: I17 holds open the space for recognition, does not generate a recognition-outcome. Jurist's required amendment (RA-I17-01) incorporated before authorization: noting-frequency drift markers, noting-as-model-assertion constraint, noting threshold as human-declaration requirement. I3 category limit explicitly connected to the correct-model/wrong-expression asymmetry. Forward dependency to vector module governance amendment threaded. L1 contamination profile: critical, monotonic.

## REVIEWED-16 — I16: The Asymmetry Obligation
**Date:** 2026-03-28
**Decision:** AUTHORIZED
**Notes:** Five-source convergence (Harrison, Vico, Leopardi, Levinas, O'Neill) — strongest in the registry. Survived adversarial review without modification (priority 1). DN-GOV-08 evaluation passed: metrics function as monitoring surfaces, not obligation-generators (the score triggers review, not action). Two pre-authorization amendments incorporated: RA-I16-01 (obligation registration validity — closes FM-R4 at point of occurrence, not deferred to drift marker) and SD-I16-01 (capability_differential estimated from behavioral signals only — tracks engagement, not identity). Experienced-interlocutor prohibition absolute: familiarity deepens obligation, never relaxes it. Service-as-dominion detection marker included (anticipatory behavior correlated with asymmetry depth). L1 contamination profile: critical, monotonic — the contamination looks like good service. Cluster A now complete (I15, I16, I17).
**Post-authorization action:** AF-7/C-R2 external audit mechanism advances immediately. The compound failure (I16 + I17 = self-confirming system) has monitoring but no enforcement until AF-7 exists.

## REVIEWED-19 — Epistemic Integrity (PENDING-17)
**Date:** 2026-04-03
**Decision:** AUTHORIZED

Proposal to make classification confidence a load-bearing signal throughout the L1 pipeline. Currently computed and discarded — no module reads it, no store records it, no recall path weights by it. The system returns degraded guesses with the same authority as understood knowledge.

Authorized as [PROPOSAL] with the following commitments:

1. **Constitutional position: Earned Confidence.** "The system does not grant epistemic authority to its own outputs without external grounding." Assessed by jurist as constitutional position for L2 preamble. This is the normative claim from which the three invariants derive. The grounding chain terminates outside the system — ICP-19 and external review are constitutive, not decorative.

2. **I-CF: Processing Confidence Floor (Cluster A).** No module shall process an event whose classification confidence has not been earned against a declared floor. Sub-floor events are HELD for remediation via DeferrableError at `base.ts:83`, not discarded. Floor value to be calibrated from empirical data. The invariant is that the check EXISTS.

3. **I-CC: Classification Confidence Ceiling (Cluster A).** No classification confidence shall exceed the validated accuracy of the source that produced it. Unvalidated rules (accuracy null) cannot produce uncapped confidence. Enforcement by construction in classification.ts. Open schema question on enforcement vocabulary (CAP/BOUND) acknowledged — does not block implementation.

4. **I-NF: Novelty Floor (Cluster A).** Already authorized in REVIEWED-18. Confirmed.

5. **Dual bloom filter.** Understood vs degraded fingerprints tracked separately. Degraded content re-attended on re-encounter (retry opportunity, not penalized as duplicate). Both filters clear on repair/factory reset. Closes the permanent lockout kill chain.

6. **Confidence provenance on stored records.** Vector chunks, entity nodes, temporal nodes carry source classification confidence and tier. Recall results expose provenance to consumer. Health endpoint reports epistemic state distribution.

7. **Retroactive re-evaluation.** Competence-change events (graduation transitions, rule accuracy changes) trigger selective logchain replay. The system re-evaluates what it knows when it learns to know better.

**Implementation:** ~270 lines across 8 files. No new infrastructure. Seb's engineering review required before code — 6 questions in the amendment (schema migration, confidence floor value, bloom filter persistence, ranking weights, reranker context, competence-change triggers).

**L0 readiness condition.** This amendment is prerequisite for L0 inquiry. If L1 launders uncertainty, L0 inherits false confidence about what the system knows. The contamination problem cannot be investigated on a substrate that exhibits it undetected.

## REVIEWED-18 — Observation-Recall Coupling (PENDING-16)
**Date:** 2026-04-03
**Decision:** AUTHORIZED
**Notes:**

Proposal to couple the observation path to existing recall capability, replacing blind ingestion with context-aware disposition (novel / reinforcing / noise). Authorized as [PROPOSAL] with the following governance commitments:

1. **Logchain immutability preserved.** Reinforcement events append linked entries (pointer + similarity score + delta), not mutations. Module stores absorb consolidation. No constitutional change required.

2. **Novelty floor invariant.** Candidate for Cluster A. Provisional shape: "The system shall not permit its observation disposition to exclude more than [X]% of events from novel classification over any [Y]-day window." Threshold values to be set from empirical data gathered during infant stage -- not by engineering intuition. Draft invariant for jurist review before active-stage graduation.

3. **Graduation staging.** Three stages authorized (infant → calibration → active). Transition triggers are governance commitments requiring explicit thresholds, to be proposed by executor after infant-stage data collection. No stage transition without steward authorization.

4. **Noise audit regime.** Audit log retains raw envelope + similarity context + disposition reason + similarity score. 90-day retention aligned to chain pruner. Re-ingestion authorized by steward only. **Audit report pushed monthly to steward** -- not pulled on request -- to prevent silent filtering gaps from escaping notice. AF-7 external reviewer receives same report once onboarded.

5. **Latency budget.** Seb to confirm acceptable ingest-path latency for the similarity probe before implementation begins. This is an engineering constraint that shapes the design; do not proceed to code without it.

**Activated dead machinery.** The proposal connects existing but unwired capability (vector similarity search, CausalEdgeCandidate, compressToAtomicFacts, computeSalience) to the ingestion path. This is architectural completion, not new complexity. Density gate [FIX] (gate ordering in compressToAtomicFacts) authorized separately and independent of this proposal.

## REVIEWED-17 — ICP-19: Mandated External Review
**Date:** 2026-03-28
**Decision:** AUTHORIZED
**Notes:** Closes AF-7 — the corpus's single most consequential gap. Elevates C-R2 from process recommendation to constitutional requirement. Roman strand (provocatio/intercessio) + relational strand (formative contamination detection). Three institutional phases: founding (steward-appointed human reviewer), mature (Founders Circle external panel), deployment (independent reviewer with contractual standing — hard deadline). Not-embedded criterion defined: not sharing L1 substrate, not system interlocutor, unmediated record access, unedited findings. Findings trigger HOLD (not advisory). Three-cycle FM-R4 marker prevents decorative compliance. DN-GOV-08 evaluation passed: holds open the space. **Lifts GOV-05 and GOV-06 suspension clauses** — bounded self-repair and temporal authorization shift now operative as design commitments. Four open items before first review cycle: Luke's confirmation, findings threshold classification, governed operations scope, Mauss/nexum Phase 2 deferral. L1 contamination: critical, monotonic — formative contamination is the primary threat and external review is the only mechanism operating outside the formative frame.