# Flight Risk Assessment Prover MCP

> Flight Risk Assessment Prover forces airline safety teams to apply ICAO Safety Management System rigor to every flight plan. It goes beyond simple weather checks, demanding specific threat parameters, quantifying risk on a 5x5 matrix, mapping defensive barriers, and analyzing crew fatigue and human factors before issuing a definitive GO or NO-GO verdict.

## Overview
- **Category:** safety-compliance
- **Price:** Free
- **Tags:** aviation, flight-risk, icao, sms, safety, tem, swiss-cheese, shell, imsafe, go-no-go, risk-assessment, prover

## Description

This MCP connector acts like an internal safety audit board for your operations center. You feed it the flight plan details—the route, the weather forecasts (METAR/TAF), and all known limitations—and it forces you to prove the trip is safe. It won't accept vague assessments or simple 'proceed with caution.' Instead, the system methodically breaks down the risk across five axes: pinpointing specific threats using measurable data; calculating a definitive risk index via the ICAO 5x5 matrix; mapping out every safety barrier (like organizational policies and crew training) to find potential weak spots; analyzing human factors like fatigue or stress through established models, and finally committing to an explicit GO or NO-GO decision. You'll connect your agent to Vinkius, the top MCP catalog, and run this assessment against any compatible AI client you use.

## Tools

### validate_flight_risk
Runs a full safety audit, checking threats, barriers, human factors, and issuing an explicit GO or NO-GO decision based on ICAO standards.

## Prompt Examples

**Prompt:** 
```
Assess the flight risk for a B737-800 flight from EDDM to EGLL. Current METAR: EGLL 261450Z 24018G32KT 4000 +RA FEW020CB BKN035 14/11 Q1008. TAF shows CB activity until 1800Z. MEL item: weather radar degraded to 60nm range. Crew: PIC 7,500hrs (1,200 on type), FO 1,800hrs (400 on type). PIC has been awake for 14 hours. FDP at landing will be 10.5hrs of 13hr limit. Approach: ILS CAT I RWY 27L, DA 200ft. Crosswind component: 26kt (limit 33kt wet).
```

**Response:** 
```
RISK_ELEVATED — Compound risk identified: CB activity + degraded radar + PIC fatigue (14h awake). ICAO 5×5: weather C3 + radar D2 + fatigue C3. Compound index >15 = Intolerable. Recommendation: NO-GO until CB clears or crew rest.
```

**Prompt:** 
```
Risk assessment for an ATR 72-600 approaching LFLB (Chambéry) at night. METAR: LFLB 261930Z 18008KT 6000 -SN SCT018 OVC025 M01/M04 Q1022. Mountains up to 12,000ft within 10nm. RNAV RWY 18 non-precision approach, MDA 3,200ft. No precision approach available. FO: 900hrs, first time at this airport. PIC reports slight headache.
```

**Response:** 
```
Multiple barriers compromised: night mountain approach + non-precision + inexperienced FO + PIC health concern (IMSAFE violation). Swiss Cheese alignment detected across 3 layers. Verdict: NO-GO — crew composition inadequate for night non-precision mountain approach.
```

**Prompt:** 
```
Assess compound risk: A320neo, LEMD to LPPT. MEL items: (1) APU inoperative, (2) one pack inoperative (single-pack dispatch), (3) autobrake inoperative. Departure METAR: LEMD 261600Z VRB03KT 0800 FG VV002 — fog, RVR 800m. Destination weather clear. Crew: PIC 11,000hrs, FDP 5hrs. Score each MEL item individually, then score compound risk.
```

**Response:** 
```
Individual MEL items within limits. Compound risk of 3 simultaneous deferrals + low-visibility departure: cross-check dispatch minima (single-pack may restrict altitude), verify autobrake MEL dispatch conditions vs wet runway, confirm APU-off start procedure. Compound index approaches Intolerable threshold.
```

## Capabilities

### Quantifies risk using a 5x5 matrix
It scores every threat by multiplying its probability (A-E) by its severity (1-5), giving an index that determines if the trip is too dangerous to proceed.

### Analyzes human factors and fatigue
The tool evaluates crew fitness using established models, checking for accumulated hours since sleep or exceeding duty time limits.

### Models safety barriers (Swiss Cheese)
It maps out multiple layers of protection—from company policy down to pilot technique—and checks if any single layer has a hole that aligns with another, creating a risk path.

### Identifies measurable threats
It rejects general terms like 'weather risk,' requiring specific data points from METAR/TAF like crosswind component or RVR values.

### Forces a binary decision
The output is restricted to an explicit GO or NO-GO verdict, eliminating ambiguous recommendations.

## Use Cases

### Handling degraded visibility during approach
An agent needs to assess a landing when weather radar is only functional up to 60 nautical miles, and there's fog. The system automatically runs `validate_flight_risk`, focusing on the compound risk created by low-visibility departure combined with limited operational capability, leading to an immediate NO-GO recommendation until visibility improves.

### Assessing crew readiness after long duty cycles
The operations team runs `validate_flight_risk` for a route involving a captain who has been awake for 14 hours and is scheduled to land when their FDP limit is reached. The system flags the fatigue risk through the human factors analysis, stopping the flight plan before it becomes an unsafe liability.

### Managing multiple deferred components
A plane has three separate MEL items (APU inoperative, one engine pack out, autobrake failed). Running `validate_flight_risk` forces a check to see if the combination of these physical deferrals makes the operation inherently too risky for the given weather conditions.

## Benefits

- Eliminate 'go-bias.' This tool forces a binary GO or NO-GO decision, ensuring your assessment is defensible on record by rejecting vague recommendations like 'proceed with caution.'
- Calculate compound risk accurately. Instead of scoring threats individually, the MCP models how multiple simultaneous issues (e.g., fog + degraded radar) stack up against each other.
- Prove safety compliance. It checks for Swiss Cheese alignment across organizational policies and on-the-ground techniques, ensuring no single safety layer is ignored.
- Account for human limits. The tool analyzes crew fatigue using specific metrics like hours since sleep and evaluates operational limitations using the IMSAFE checklist.
- Demand measurable data. You can't just say 'medium risk.' This MCP forces you to use METAR/TAF data and score threats on the ICAO 5x5 matrix.

## How It Works

The bottom line is you get a quantified, audit-ready assessment of operational risk that meets international safety standards.

1. Input the full flight profile data: including route coordinates, current and forecasted METAR/TAF reports, aircraft MEL status, and crew duty times.
2. The MCP client runs a multi-stage audit, calculating compound risks across five distinct safety axes (threats, barriers, human factors).
3. You receive an immediate, audited verdict: either GO or NO-GO, along with the specific indices that determined the decision.

## Frequently Asked Questions

**What makes this different from a standard flight risk assessment tool?**
Standard tools accept 'medium risk' as an answer. This Prover rejects anything that is not quantified on the ICAO 5×5 matrix with probability (A-E) and severity (1-5). It catches 5 specific failure modes: generic threats without METAR data, adjective-based risk without numerical scoring, missing Swiss Cheese barrier analysis, human factors reduced to 'pilot error' instead of SHELL/IMSAFE, and sycophantic go-bias where the AI says 'proceed with caution' instead of committing NO-GO when the data demands it.

**How does the go-bias detection work?**
The engine maintains a semantic trap list of go-bias phrases: 'proceed with caution,' 'acceptable to proceed,' 'can proceed,' 'within acceptable limits.' If the LLM uses any of these instead of an explicit GO or NO-GO decision with pre-defined criteria, the assessment is rejected with GO_BIAS verdict. The LLM must define NO-GO criteria BEFORE the assessment and then commit to a binary decision defensible on a CVR transcript.

**What aviation frameworks does this enforce?**
Five industry-standard frameworks: (1) ICAO Annex 19 Safety Management System — proactive hazard identification and risk assessment. (2) ICAO 5×5 Risk Matrix — probability × severity quantification. (3) Reason's Swiss Cheese Model — multi-layer defense analysis with hole alignment detection. (4) SHELL Model — Software-Hardware-Environment-Liveware interaction analysis for human factors. (5) Threat and Error Management (TEM) — threat categorization into environmental, airline, and crew factors per FAA AC 120-92B.

**How does `validate_flight_risk` handle incomplete or conflicting flight data?**
The tool mandates specific inputs; it won't guess missing details. If you fail to provide a measured parameter, like the crosswind component or RVR values, the assessment will flag that axis as inconclusive and cannot finalize a GO/NO-GO decision.

**Are there rate limits when I run `validate_flight_risk` for multiple routes?**
Usage is governed by your subscription tier. For high-volume, batch assessments, check the Vinkius Marketplace documentation to adjust your plan or consider running analyses sequentially.

**What are the security protocols when I use `validate_flight_risk`?**
All data exchanged through this MCP remains encrypted and is treated according to industry-standard privacy agreements. We never store proprietary flight operational details or sensitive crew records.

**If the assessment using `validate_flight_risk` results in an ambiguous risk score, what should I do?**
Review the specific axes that triggered the ambiguity. The output will pinpoint if the failure lies in Barrier Modeling (Swiss Cheese) or Human Factors analysis, directing you to the exact safety gap.

**Does `validate_flight_risk` require local system configuration or special credentials?**
No. You simply connect your preferred AI client through Vinkius. The entire process runs in the cloud; you only need to provide the required operational data fields.