How does the BMAD Risk Scoring formula work?

It calculates a Risk Score by multiplying Probability (0.0 to 1.0) by Impact (1 to 10). Scores above 9.0 are considered Critical and trigger a FAIL decision at the quality gate.

Does this skill work with any repository?

It is designed to work within the Specweave framework, specifically evaluating increments located in the .specweave directory of your project.

How do I activate the Quality Judge in Claude Code?

The skill auto-activates when you use natural language phrases like 'assess quality' or specific commands like '/sw:qa {increment-id}'.

What is the LLM-as-Judge pattern used in this skill?

LLM-as-Judge is a technique where a large language model evaluates outputs using structured reasoning, chain-of-thought processing, and established criteria to provide consistent, human-like quality assessments.

What is the difference between a PASS and CONCERNS decision?

A PASS indicates the increment is ready for production. A CONCERNS decision means high risks or quality issues were found that should be addressed, but aren't necessarily blockers.

Increment Quality Judge

Name: Increment Quality Judge
Author: anton-abyzov

byanton-abyzov

•

Security & Testing

Evaluates software specifications and plans using AI-powered risk scoring and formal quality gate validation.

About

The Increment Quality Judge v2.0 implements the LLM-as-Judge pattern to provide rigorous, automated quality assessments for software development increments. By evaluating seven key dimensions—including clarity, testability, and a specialized BMAD-pattern risk score—it establishes formal quality gates (PASS/CONCERNS/FAIL) to prevent regressions and security vulnerabilities before implementation begins. This skill integrates directly into the Specweave framework, ensuring that AI-generated specs, tests, and docs meet production-grade standards through objective, evidence-based reasoning and actionable mitigation recommendations.

Key Features

LLM-as-Judge pattern for consistent, reasoning-based quality evaluations
13 GitHub stars
Seven-dimension analysis covering clarity, testability, completeness, and feasibility
BMAD Risk Scoring (Probability × Impact) to identify and categorize critical vulnerabilities
Automated NFR checking for security, performance, and scalability requirements
Formal Quality Gate decisions with PASS, CONCERNS, and FAIL status levels

Use Cases

Pre-implementation validation of software specifications to identify logic flaws early
Systematic risk assessment of architectural plans to mitigate security and technical debt
Automated quality gate checks to ensure increments meet production readiness standards

About

Key Features

LLM-as-Judge pattern for consistent, reasoning-based quality evaluations
13 GitHub stars
Seven-dimension analysis covering clarity, testability, completeness, and feasibility
BMAD Risk Scoring (Probability × Impact) to identify and categorize critical vulnerabilities
Automated NFR checking for security, performance, and scalability requirements
Formal Quality Gate decisions with PASS, CONCERNS, and FAIL status levels

Use Cases

Pre-implementation validation of software specifications to identify logic flaws early
Systematic risk assessment of architectural plans to mitigate security and technical debt
Automated quality gate checks to ensure increments meet production readiness standards