This is a really useful way to turn a suspicion into something you can actually look at. The second-feed comparison is the key part, because without it every stop that gets taken at the high of a spike feels personal.
One thing I'd add to the method is recording the spread at the moment the stop triggered, not just the fill price. On a buy position your stop is hit on the bid, but most charts show the bid only, so a long stop behaves the way you'd expect. On a sell position the stop is triggered by the ask, and if the spread jumps from 0.2 to 2 pips at rollover, the ask can touch your stop while the bid line on the chart never gets near it. That looks exactly like a stop hunt on the chart, but it's really a spread event. Splitting the log into long stops and short stops can show this quite clearly, because a broker problem would hit both sides while a spread problem mostly shows up on shorts.
The other useful check is distance. If the stops that look suspicious are mostly the tight ones, just a few pips from entry, then the issue is usually that the stop was inside normal noise for that time of day rather than anyone targeting it.
If the pattern still holds after accounting for spread and distance, and it only happens on one broker's feed, then it's worth raising with the broker with the data in hand. That conversation goes much better with a spreadsheet than with a screenshot.