KB-3645

T1 FIX7 Focused Review - 03 14 Readiness Gate Adapter Review (SUPERTRACK C)

4 min read Revision 1
QT001FIX7T1readinesssupertrack-c

03 - 14 Readiness Gate Adapter Review (SUPERTRACK C)

Source: artifact 03 (full read, content_length 2178). Verdict: READINESS_GATES_NEED_CORRECTION (minor; one blocking item CP-05).

Count check - PASS

Exactly 14 gates: OBJECT_AUTHORITY_IMMUTABLE, READINESS_MANIFEST_EXACT, SIGNOFF_AUTHENTIC, CAPABILITY_BEHAVIORAL, DEPENDENCY_TRUTH, NO_BYPASS_ALL_BLOCKED, CANONICAL_HASH_SHA256, PLANNER_NOT_CLONE, SCALE_SAFE, GATEWAY_PASS, FREEZE_PASS, PERMIT_POLICY_PASS, TIER_POLICY_PASS, WRITER_FAIL_CLOSED.

Verified strong (each required element present)

  • Uniform adapter signature: (control_epoch)->(gate_code,value_boolean,observed_at,evidence_id,source_set_sha256), STABLE read-only, source/signature/dependency-bound, exactly one row.
  • Missing/extra/NULL behavior: "Zero/multi/NULL/wrong/stale/unknown=false" - fail-closed on every non-clean state.
  • Exact-set enforcement: "both-EXCEPT" (expected-minus-actual AND actual-minus-expected) and "no bool_and-alone" - directly answers the prior NULL-ignore and missing-gate failure modes.
  • Per-gate data source named (fact_object_authority, fact_readiness_manifest_exact, fact_signoff_authentic, ... fact_writer_fail_closed).
  • Per-gate freshness explicit: 300s (most), 900s (signoff/capability/scale), 86400s with immediate source-drift invalidation (dependency).
  • Per-gate negative evidence named (e.g. SIGNOFF_AUTHENTIC fake/expired/self/same-human=false; READINESS_MANIFEST_EXACT delete-add-null=false; CANONICAL_HASH key-mutation=false).
  • Each item binds adapter/source/signature/source-set/rule/freshness/negative-evidence/epoch into the readiness hash. "Only one typed TRUE with fresh immutable evidence is acceptable."
  • READINESS_MANIFEST_EXACT is self-sealing (one active sealed exact-14, count/hash/ordinal) - the gate set cannot be silently shrunk to green; NO_BYPASS_ALL_BLOCKED separately seals the 14 vectors.

This is a strong satisfaction of prior correction #2.

Blocking gap

C-1 Gate thresholds read as adapter literals, not sealed manifest values (CP-05)

SCALE_SAFE states "perf <=600000ms <=1073741824 bytes zero errors"; CAPABILITY_BEHAVIORAL states "exact 3 fresh behavioral runs"; READINESS_MANIFEST_EXACT and CANONICAL_HASH_SHA256 reference the constants 14 and 7. For zero-hardcode, every such denominator/threshold must be read from a sealed manifest row (SCALE_SAFE -> workload_profile_manifest #23 / capability_measurement_requirement #14 operator+operand; "3 runs" -> capability_manifest workload / measurement requirement; "14" -> readiness_gate_manifest.expected_item_count; "7" -> hash_component_manifest contract count) - never embedded as an integer in the adapter SQL. The artifact does not state this binding, so as written the adapter could legitimately be authored with literals. That is disguised hardcode. Codex must state, per gate, that the threshold/denominator is sourced from the named sealed manifest field and that the adapter contains no numeric policy literal.

Minor notes (non-blocking, fold into CP-05)

  • State that the readiness denominator used in the exact-set comparison is manifest.expected_item_count of the readiness_gate_manifest (so "14" is data, not literal). Same for the "7" used by CANONICAL_HASH_SHA256 (hash_component contract count).
  • DEPENDENCY_TRUTH 86400s with "immediate drift" is the correct pattern (age cap AND instant invalidation on source-hash change) - keep.

Cross-layer coherence - PASS

14 gates != 14 bypass vectors (NO_BYPASS_ALL_BLOCKED is one gate consuming all 14 vectors) - no conflict. Gates feed readiness hash H03/H01/H02 consistently with doc 04.

Verdict

READINESS_GATES_NEED_CORRECTION. Gate set, formulas, sources, freshness, NULL/missing/extra behavior, both-EXCEPT, no-bool_and-alone, and self-seal are all present and correct. The single correction is to seal all thresholds/denominators as manifest values with an explicit "no numeric policy literal in adapter" rule (CP-05).

Back to Knowledge Hub knowledge/dev/reports/architecture/t1-fix7-spec-artifact-focused-review-and-correction-proposal-2026-06-07/03-14-readiness-gate-adapter-review.md