Raw log files are huge, repetitive and full of secrets. Cleaning them first means 80% fewer tokens and no leaked API keys. All processing happens in your browser.
Lines—
Errors—
Warnings—
Repeated—
Est. tokens—
Why send raw logs to an AI at all?
A typical error investigation pastes the whole log file into a chat. That is the worst version of the data you have:
Retry loops repeat identical lines thousands of times — pure token waste.
Timestamps and request IDs on every line double the size again.
Authorization headers, tokens and emails sit in plain sight.
What "Clean for AI" does
Collapses repeated runs — 3 copies kept, the rest replaced with a [repeated Nx] marker.