Closes the loop from the AI UX Auditor's findings to action. It consumes the UX Auditor's audit (the recurring UX issues, with their prevalence and severity) and turns each into a concrete A/B experiment proposal — hypothesis, exactly what to change, what to measure, expected impact and effort — ranked by prevalence x severity, with quick-wins (high impact, low effort) surfaced first. Each proposal can be tracked (proposed / planned / running / done / dismissed). Requires the AI UX Auditor plugin; if it is not enabled the engine says so rather than inventing experiments. Transparent mapping (not an LLM). Reads existing data — no extra tracking, cookieless. Feeds the Experiments panel of the Visitor Intelligence hub.