Flight Test Data Anomaly Review Agent

Each test flight produces hours of sensor data, and a slow trend can hide among thousands of channels. After each flight, this agent loads the data and checks it for gaps and bad sensors. If files are missing or channels are bad, it requests the files and marks those channels before going on. It then compares every parameter with its limits and with earlier flights at the same test point. It flags exceedances and unexpected trends, such as a slowly rising temperature. Single-sample spikes are checked against the sensor's known noise first. For each anomaly it pulls matching pilot notes and video timestamps. It then checks whether each flag could be a sensor fault and marks those as data issues. The engineer decides what needs investigation. Edge case: a one-sample spike within sensor noise is not called an exceedance.