What problem does it solve? Auditing a mature Home Assistant setup by hand is slow and error-prone, and naive automated reviews re-flag dozens of already-settled design decisions. This Skill runs a disciplined, read-only review of the Home Assistant domain that filters out known accepted trade-offs and adversarially verifies findings before reporting them. ## Core Features & Use Cases - Scoped read-only audit: Dispatches a reviewer agent over the home-assistant Ansible role covering automations, scenes, scripts, template sensors, and configuration, with explicit instructions to make no changes. - Memory-primed signal boosting: Loads prior review-state and accepted-decision memories first so settled decisions are not re-flagged as new findings. - Adversarial verification: Every High/Medium finding is independently checked by a skeptic agent against role documentation, sanctioned writers, and live state before reaching the report. - Use Case: Ask for a review of your Home Assistant setup and receive a prioritized High/Medium/Low report with file:line citations, tagged as gaps, improvements, or additions, plus a top-priorities shortlist. ## Quick Start Ask the assistant to run a read-only review of the Home Assistant setup and report prioritized findings without making any changes.
