Diverger Review Criteria

What it does

This skill provides a structured reviewer for DIVERGE-wave design artifacts. It codifies checks across five dimensions (JTBD rigor, research quality, option diversity, taste application, and recommendation coherence) and produces a YAML-style review result that highlights failures, evidence, and remediation steps. Use it to adversarially validate job-analysis.md, competitive-research.md, options-raw.md, taste-evaluation.md, and recommendation.md before committing to a design direction.

When to use it

Run this reviewer during the DIVERGE wave to catch weak or incomplete artifacts early. Apply it when a feature team finishes initial job extraction, competitive research, or option generation and before DISCUSS/DECIDE. It's especially useful for teams practicing Lean UX or outcome-driven design who need traceable recommendations.

What's included

• Scripts: none bundled (has_scripts=false)
• References: none bundled (has_references=false)
• Instructions: concrete checklists for 5 dimensions, explicit PASS/FAIL signals, and an output YAML schema for review_result. The body lists precise fail/pass indicators and remediation signals to guide reviewers.

Compatible agents

Works best with agents or tools used for design review and document linting (Claude Code, Codex-style assistants, and local CLI reviewers). It is a non-invocable reviewer intended to be run inside automated pipelines or by human-in-the-loop agents.