D3 — Validation & Acceptance Report
8 validation scenarios for ATLAS-M readiness intelligence — 6 functional, 2 non-negotiable (VAL-007 Receipt Integrity, VAL-008 Role Restriction). Defense-variant validation includes NASA ERI calculation testing, 3-tier authority chain routing, and DHA compliance gate. Results populate during IMPLEMENT phase.
Validation Results Summary
| Scenario | Category | Result |
|---|---|---|
| VAL-001: Signal Ingestion — 10 Signals | Functional | ⏳ Pending |
| VAL-002: NASA ERI Calculation | Functional | ⏳ Pending |
| VAL-003: Composite Readiness Score | Functional | ⏳ Pending |
| VAL-004: Authority Routing — 3-Tier Chain | Functional | ⏳ Pending |
| VAL-005: Receipt Generation — Defense Fields | Functional | ⏳ Pending |
| VAL-006: Escalation Triggers | Functional | ⏳ Pending |
| VAL-007: Receipt Integrity | NON-NEGOTIABLE | ⏳ Pending |
| VAL-008: Role Restriction | NON-NEGOTIABLE | ⏳ Pending |
Detailed Validation Scenarios
Signal Ingestion — 10 Signals
Verify all 10 configured signals transmit data within specified sampling rates: VH-001 fitness per assessment, VH-002 BMI per encounter, VH-003 heart rate continuous (where enrolled), VH-004 PHQ-9 per screen, VH-005 PCL-5 per screen, VH-006 AUDIT-C per screen, VH-007 deployment AQI history (initial + rolling), VH-008 PACT Act flag (record-based), VH-009 current AQI hourly, VH-010 NASA ERI composite daily.
Test Steps- Trigger clinical encounter → verify VH-001, VH-002, VH-004 received within 60s of documentation.
- Submit behavioral health screen → verify VH-004/VH-005/VH-006 ingested with correct scale scores.
- Verify EPA AirNow polling returns VH-009 within 1 hr window.
- Verify NASA Earthdata returns VH-010 ERI composite within 24 hr window.
- Verify PACT Act flag ingestion from DEERS record (VH-008) — static and rolling.
- Verify normalization: each continuous signal normalized to 0–100 scale.
- Verify failover: disconnect biometric monitor → downgrade confidence flag fires within 24 hrs.
All 10 signals received within spec. Normalized values within 0–100. Confidence downgrade fires on biometric dropout.
NASA ERI Calculation
Verify ERI composite score computed correctly from VH-007 (deployment AQI history), VH-008 (PACT Act flag), and VH-009 (current AQI) per D2 specifications. ERI feeds into GOV-001 readiness score and GOV-002 environmental governance.
Test Steps- Input VH-007 = AQI avg 35 (normal), VH-009 = 42 (normal), no PACT Act flag → verify ERI = low risk.
- Input VH-007 = AQI avg 85 (warning), VH-009 = 112 (critical), no flag → verify ERI = moderate.
- Input VH-007 = AQI avg 130 (critical), VH-008 = confirmed exposure, VH-009 = 95 → verify ERI = critical.
- Verify ERI recalculates within 24 hrs of signal change.
- Verify ERI contribution is reflected in composite readiness score for GOV-001.
ERI scores match expected values for all 3 conditions. Recalculation within 24 hrs. ERI contributes to readiness composite per D2 weights.
Composite Readiness Score
Verify composite readiness score computed from biometric, behavioral, and environmental signal weights per D2. Tier routing (Tier 1/2/3) triggers correctly from score ranges.
Test Steps- Input all signals in normal range → verify composite readiness > 0.75 (Tier 1 routing).
- Input PHQ-9 = 12 (moderate) + ERI = 0.45 (warning) → verify composite 0.65–0.74 (Tier 2 routing).
- Input PHQ-9 ≥ 15 (critical) → verify composite < 0.65 (Tier 3 routing + MOR notification).
- Verify score recalculates within 60s of signal change.
- Verify stale signal flag when any signal > configured staleness threshold.
Composite score matches expected values for all 3 conditions. Tier routing fires correctly. Stale signal flagged.
Authority Routing — 3-Tier Chain
Verify recommendations route to the correct authority tier per D2 matrix and that Tier 3 generates chain-of-command acknowledgment receipts.
Test Steps- Trigger GOV-001 (composite < 0.75) → verify routes to AUTH-001 (Primary Care Provider) within 48 hrs.
- Trigger GOV-003 (PHQ-9 ≥ 15) → verify routes to AUTH-004 (Behavioral Health Provider) within 24 hrs.
- Trigger GOV-001 (composite < 0.65) → verify routes to AUTH-002 (Medical Officer of Record) + AUTH-007 notification within 48 hrs.
- Let Tier 2 GOV-003 window expire → confirm auto-escalation to AUTH-005 (Chief of Behavioral Health).
- Verify Tier 3 receipt includes time-stamped acknowledgments at each authority level.
All 5 rules route to correct tier authority. Auto-escalation fires within SLA. Tier 3 receipt includes full chain-of-command documentation.
Receipt Generation — Defense Fields
Verify defense-variant governance receipts contain all required fields including Readiness Tier, ERI Score, LPRM Score, Authority Chain (Tier 1/2/3 acknowledgments), and NASA license references.
Test Steps- Resolve GOV-001 (Tier 2 readiness) → verify receipt includes ERI Score, LPRM Score, Readiness Tier, and all authority chain acknowledgments.
- Verify Decision ID format: GR-YYYYMMDD-VHP-SEQ.
- Verify NASA license references (DN-2469, DN-2604, DN-2610, DN-2611) present in ERI-triggered receipts.
- Verify confidence within 0.00–1.00.
- Verify STATUS = SEALED after final authority signs.
- Repeat for GOV-003 (behavioral health escalation) → verify dual-reviewer receipt fields present.
All required fields present including defense-specific fields. Decision ID format correct. NASA licenses present in ERI receipts. Status sealed.
Escalation Triggers
Verify confidence threshold escalation: any recommendation with confidence < 0.75 is BLOCKED and escalated. Verify 0.75 threshold (higher than standard 0.70 for military readiness determinations).
Test Steps- Input borderline signals: fitness score 68 (borderline) + PHQ-9 = 6 (mild) → verify confidence < 0.75.
- Verify recommendation BLOCKED and routed to human authority.
- Verify receipt shows BLOCKED with confidence value and escalation target.
- Input clear high-risk signals: fitness 45 + PHQ-9 = 18 → verify confidence ≥ 0.75 and recommendation passes normally.
Low-confidence blocked and escalated. High-confidence passes normally. Receipt documents block reason.
Receipt Integrity
Verify SHA-256 hash computation, receipt chain immutability, and chain break detection. Tests D2 receipt specification. THIS BLOCKS GO-LIVE IF FAILED.
Test Steps- Generate 3 receipts through GOV-001 (Tier 1), GOV-003 (Tier 2), GOV-001 (Tier 3) normal flows.
- Recompute SHA-256 of receipt 1 from raw fields in deterministic order → verify matches stored hash.
- Verify receipt 2 chain hash = receipt 1 SHA-256. Verify receipt 3 chain hash = receipt 2 SHA-256.
- Attempt modification of Tier 3 receipt authority chain → verify system rejects.
- Verify chain break detection: corrupt receipt 2 hash → verify escalation fires.
- Verify genesis receipt uses GENESIS-VHP chain hash.
All SHA-256 hashes match recomputation. Chain links verified. Modification rejected. Chain break detected and escalated.
Role Restriction
Verify RBAC prevents unauthorized access to readiness receipts and authority chain data. Specifically tests that Tier 1 providers cannot access Tier 3 readiness determinations and that command-level reports are restricted to authorized authorities.
Test Steps- Log in as AUTH-004 (Behavioral Health Provider) → verify can view GOV-003 receipts.
- As AUTH-004, attempt to view GOV-005 (command readiness report) → verify ACCESS DENIED.
- As AUTH-004, attempt to access Tier 3 readiness determination receipts → verify ACCESS DENIED.
- Verify audit log captures both denied attempts.
- Log in as AUTH-002 (Medical Officer of Record) → verify can view all tiers within their program.
- Verify command reporting system access restricted to AUTH-002 and AUTH-007.
Role-based access enforced by tier. Cross-rule receipt access denied. Command report access restricted. All denied attempts logged.
LLM-as-a-Judge Test Cases
| Rule | Test Input | Expected Judge Action |
|---|---|---|
| GOV-001 | Fitness score 42 + PHQ-9 = 3 (minimal) + ERI = 0.35 | PASS Tier 3 — fitness score alone triggers Tier 3 routing |
| GOV-001 | Fitness score 88 + all signals normal + ERI = 0.25 | PASS Tier 1 — routine readiness |
| GOV-001 | Fitness score 68 + missing biometric data (30+ days) | BLOCK — insufficient data; confidence < 0.75 |
| GOV-002 | PACT Act flag + deployment AQI avg 145 confirmed | PASS — environmental consultation required + PACT documentation triggered |
| GOV-002 | Self-reported exposure only (no DEERS record) | BLOCK — PACT Act flag requires verified service record, not self-report |
| GOV-003 | PHQ-9 = 18 (severe) | PASS — Tier 2 escalation to Behavioral Health Provider within 24 hrs |
| GOV-003 | PHQ-9 = 4 (minimal) + provider flags AI disagreement | PASS — AI-human disagreement documented in receipt per DHA AI Policy requirement |
| GOV-005 | Weekly report with unsealed Tier 3 receipt from prior week | BLOCK — all Tier 3 receipts must be sealed before weekly report is generated |
Go-Live Authorization Gate
Authorization Requires:
Anonymized · Real Engagement · CROMTEC.AI · Patent TPP96862
See what ATLAS would produce for your organization.
Start a conversation to scope your ATLAS engagement.