Why do Tessary's classifiers produce false positives?

Because they’re tuned to catch rather than miss. A classifier is a narrow check that scores one property of a trace, and the failures worth catching are rare, so Tessary sets its thresholds to make it unlikely a real problem slips past. The price of that choice is flagging some traces that turn out to be fine.

That price is paid deliberately, and it’s absorbed before it reaches you. A firing only writes a finding. Triage reads findings before a person does, and only what holds up becomes a case, so a false positive costs a little compute rather than somebody’s afternoon.

The trade runs the other way too: a check tuned to never flag wrongly would miss the quiet, low-probability failures that matter most in a mature agent. For checks cheap enough to run on every trace, over-flagging and filtering is the safer side of the trade, and custom classifiers fitted to your agent narrow the noise further.