A fixed TESS search gives an AU Mic event rate ratio of 1.21 for 2025 versus 2018–2020 under a Poisson model

We compare AU Mic flare events in public TESS sectors 1, 27, and 95 at one cadence of 120 seconds. A fixed search selects 17, 19, and 20 events with measured equivalent duration of at least 5 seconds in 20.292, 19.106, and 18.150 accepted days. The 2025 rate divided by the pooled 2018–2020 rate is 1.206. Its exact conditional 95% interval is 0.661–2.140 under an independent Poisson count model. The two-sided test gives p = 0.565. The data do not resolve a difference in this selected rate. Five method checks also give intervals that contain one. Sample gaps can divide one apparent flare into multiple selected events. The interval does not include this measurement effect. We inject 4,800 finite-exposure flare templates before the baseline fit. For input equivalent duration of 10 seconds, 95–99% of trials with widths of 2–10 minutes pass the measured 5-second cut. Only 70–86% pass at a width of 20 minutes. These quiet-time tests show loss and migration at the measurement cut. They do not give a correction for all physical flares. We provide the source FITS files, provenance, event tables, injection trials, figures, code, and reproduction checks. The result is a rate bound for a specified search. It does not establish a constant physical flare rate.

AU Mic · TESS · stellar flares · equivalent duration · event rates · injection recovery · selection effects · public data