Flight Risk Assessment Prover MCP Connector for Claude
A+A dispatch office cleared a flight into known CB activity with 'proceed with caution.' The crew never returned. Flight Risk Prover forces ICAO SMS-level threat identification with METAR data, 5×5 risk quantification, Swiss Cheese barrier modeling, SHELL/IMSAFE human factors analysis, and explicit GO/NO-GO commitment — eliminating the sycophantic go-bias that kills in aviation.
Flight Risk Assessment Prover enforces ICAO Safety Management System rigor across 5 axes that LLMs consistently fail:
Axis 1 — Threat Specificity. Generic 'weather risk' is rejected. Every threat must be named with METAR/TAF data, TEM categories (environmental, airline, crew), measurable parameters (crosswind component, RVR vs minimums, ceiling vs DA, icing level, fuel impact), and exposure duration.
Axis 2 — Risk Quantification. Adjective-based 'medium risk' is rejected. Each threat is scored on the ICAO 5×5 matrix: Probability (A=Extremely Improbable to E=Frequent) × Severity (1=Negligible to 5=Catastrophic). Compound risk and post-mitigation residual risk must be calculated. Index >15 is Intolerable (NO-GO).
Axis 3 — Barrier Modeling. 'Safety measures in place' is rejected. Swiss Cheese defense layers must be mapped: Organizational (SMS, policies) → Supervisory (dispatch, scheduling) → Preconditions (MEL items, crew fitness) → Acts (technique, monitoring). Holes identified in each layer, alignment checked across layers.
Axis 4 — Human Factors. 'Pilot error' is rejected. SHELL model analysis (Software-Hardware-Environment-Liveware) plus IMSAFE checklist (Illness, Medication, Stress, Alcohol, Fatigue, Eating). Fatigue quantified: hours since sleep, FDP position, WOCL exposure.
Axis 5 — GO/NO-GO. 'Proceed with caution' is rejected as sycophantic go-bias. Explicit GO or NO-GO with pre-defined criteria. CVR audit: 'Would I defend this decision in an investigation?'
Every lazy shortcut — generic threats, adjective risk, missing barriers, pilot blame, or go-bias — is caught by semantic traps and consistency gates before a RISK_PROVEN verdict is issued.
Related Connectors
Estimation Prover MCP
An AI estimated a database migration at 2 weeks. It took 11 weeks, cost $340K in delayed revenue, and left 3 engineers stuck in feature freeze. The estimate had no scope decomposition, no unknowns identified, no historical precedent, and no buffer. This tool forces granular scope breakdown, explicit unknown quantification, precedent mapping, and realistic buffer calculation before any timeline is committed.
Thermal Mass Estimator MCP
Calculate thermal lag, amplitude damping, and U-value for wall structures based on material properties.
Concurso Score Calculator MCP
Calculate final examination scores, manage stage thresholds, and estimate competition rankings for civil service exams.
Pesticide Dilution Calculator MCP
Calculate precise pesticide dilution, tank loads, and safety intervals.