Signals scout for PostHog A/B experiments. Watches running experiments for validity threats — sample ratio mismatch, contamination, exposure stalls, mid-run flag mutations — and lifecycle drift.
