What problem does it solve? When an AI assistant proposes options or recommendations without verifying the problem actually exists, users waste time reviewing invalid suggestions and lose trust. This Skill forces a mandatory self-check on every candidate option before it appears in a reply, filtering out unverified or redundant proposals. ## Core Features & Use Cases - Seven-Question Gate: Each option must pass checks for pre-verification via grep and spec reading, world-class benchmarks, internal design-system canonical alignment, single-source-of-truth enforcement, rule-of-three placement, downstream absorption, and full issue mapping. - Inline Evidence Tables: Outputs a structured markdown table showing pass/fail status per option with explicit reject reasons, so users can see the verification process. - Issue Mapping: For plan iterations, every user-raised issue is mapped to a step, marked informational, or explicitly dropped with a reason. - Use Case: Before replying with "Option A / B / C" for a design-system migration, run this gate to grep existing component usage, cite spec sections, and reject any option lacking file:line evidence. ## Quick Start Ask the assistant to propose options for a design-system change and require each option to pass the seven-question verification table with cited file and spec evidence before listing it.
