Inspect indicator

Pick a single indicator and see every comment that cites it — alias-resolved, so merged variants are included. Use this to sanity-check whether an indicator (e.g. "unnatural blinking") is being flagged correctly. The Matched column shows the raw phrase(s) the comment actually used, and Found shows whether each came from the LLM, semantic expansion, or keyword matching. You can rename the indicator or remove a comment that's obviously mis-attributed.