Skill

qa

by majiayu000

AI Summary

A quality assessment skill that runs automated validation, AI-powered evaluation, and risk scoring on SpecWeave increments to enforce quality gates. Developers and QA teams benefit from automated pass/fail decisions with detailed reasoning.

Install

# Add to your project root as SKILL.md
curl -o SKILL.md "https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/skills/data/qa/SKILL.md"

Description

Run quality assessment on a SpecWeave increment with risk scoring and quality gate decisions

Usage

`bash /sw:qa <increment-id> [options] `

/sw:qa - Quality Assessment Command

IMPORTANT: You MUST invoke the CLI specweave qa command using the Bash tool. The slash command provides guidance and orchestration only.

Purpose

Run comprehensive quality assessment on an increment using: • ✅ Gate 1: Rule-based validation (130+ automated checks) • ✅ Gate 2: LLM-as-Judge (AI quality assessment with chain-of-thought reasoning) • ✅ Gate 3: Risk scoring (BMAD Probability × Impact quantitative assessment) • ✅ Quality gate decisions (PASS/CONCERNS/FAIL)

LLM-as-Judge Pattern

This command implements the LLM-as-Judge pattern - an established AI/ML evaluation technique where an LLM evaluates outputs using structured reasoning. How it works: ` ┌─────────────────────────────────────────────────────────────┐ │ LLM-as-Judge Gate │ ├─────────────────────────────────────────────────────────────┤ │ Input: spec.md, plan.md, tasks.md │ │ │ │ Process: │ │ 1. Chain-of-thought analysis (7 dimensions) │ │ 2. Evidence-based scoring (0-100 per dimension) │ │ 3. Risk identification (BMAD P×I formula) │ │ 4. Formal verdict (PASS/CONCERNS/FAIL) │ │ │ │ Output: Structured quality report with: │ │ - Blockers (MUST fix) │ │ - Concerns (SHOULD fix) │ │ - Recommendations (NICE to fix) │ └─────────────────────────────────────────────────────────────┘ ` Why LLM-as-Judge? • Consistency: Applies uniform evaluation criteria • Depth: Catches nuanced issues humans might miss • Speed: ~30 seconds vs hours of manual review • Documented reasoning: Explains WHY something is an issue

Quality Score

C

Acceptable

74/100

Standard Compliance45
Documentation Quality72
Usefulness78
Maintenance Signal100
Community Signal92
Scored Yesterday

GitHub Signals

Stars100
Forks15
Issues1
UpdatedYesterday
View on GitHub

Trust & Transparency

Open Source — MIT

Source code publicly auditable

Verified Open Source

Hosted on GitHub — publicly auditable

Actively Maintained

Last commit Yesterday

100 stars — Growing Community

15 forks

My Fox Den

Community Rating

Works With

Claude Code