Verification & Quality Assurance

Score code truth and rollback changes failing quality thresholds.

67.5k|8.1k|Updated Jun 2, 2025
One-click install
npx skills add https://github.com/ruvnet/claude-flow --skill verification-quality-assurance-ruvnet
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Verification & Quality Assurance
Source: https://github.com/ruvnet/claude-flow/tree/main/.claude/skills/verification-quality
Command: npx skills add https://github.com/ruvnet/claude-flow --skill verification-quality-assurance-ruvnet

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires claude-flow@alpha, git, node, and includes references (resource) and assets (resource) components.

What problem does it solve?

This Skill ensures high code quality and reliability by providing comprehensive truth scoring, automated verification checks, and an automatic rollback system. It prevents low-quality or incorrect code from persisting, reducing technical debt and improving overall system stability, allowing you to ship with confidence.

Core Features & Use Cases

  • Truth Scoring: Real-time reliability metrics (0.0-1.0 scale) for code, agents, and tasks, providing objective quality assessment.
  • Automated Verification Checks: Validate code correctness, security, performance, and best practices across your codebase.
  • Automatic Rollback: Instantly revert changes that fail to meet a defined quality threshold (default: 0.95), preventing regressions.
  • Use Case: An AI agent commits code that introduces a critical security vulnerability. This skill can automatically detect the issue during pre-commit verification, assign a low truth score, and trigger an immediate rollback to the last known good state, preventing the vulnerability from reaching production.

Quick Start

View current truth scores

npx claude-flow@alpha truth

Run comprehensive verification check

npx claude-flow@alpha verify check

Rollback last failed verification

npx claude-flow@alpha verify rollback --last-good

Frequently Asked Questions about Verification & Quality Assurance

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically detect and prevent low-quality code from reaching production?

Automated verification assigns a truth score (0.0–1.0) to code and outputs, triggering instant rollback when quality falls below a threshold (default 0.95). This catches correctness, security, and performance issues before deployment, preventing regressions and technical debt.

Can I integrate code quality checks into my CI/CD pipeline?

Yes. This Skill runs comprehensive verification checks within CI/CD workflows, assessing code against correctness, security, performance, and best-practice metrics. Failed verifications automatically revert changes to the last known good state.

What does truth scoring measure, and how is it used?

Truth scoring provides real-time reliability metrics on a 0.0–1.0 scale for code, agents, and tasks. It quantifies quality objectively, enabling automated rollback decisions and giving teams concrete visibility into output reliability before changes persist.

Do I need additional setup to enable automatic rollback when verification fails?

No. Rollback is configured with a default 0.95 threshold and integrates with git. When verification fails, the Skill instantly reverts to the last passing commit, requiring only git and node dependencies already listed in the Skill.

How can I view current quality scores and rollback history?

Use CLI commands to inspect truth scores in real time and access rollback logs. The Skill exports live dashboards and CI/CD metrics, letting you monitor code reliability trends and review automated rollback decisions across your codebase.

What happens when an AI agent commits code with a critical security issue?

Pre-commit verification detects the vulnerability, assigns a low truth score, and triggers automatic rollback to prevent the defect from reaching production. This guards against agent-generated regressions without manual intervention.