What is a Truth Score in this skill?

A Truth Score is a real-time reliability metric scaled from 0.0 to 1.0 that evaluates the quality, security, and correctness of code, agent outputs, and completed tasks.

Can I use this skill in my existing CI/CD pipeline?

Yes, the skill provides structured JSON, CSV, and HTML export capabilities, making it easy to integrate into GitHub Actions, GitLab CI, and other automation platforms.

What specific criteria does the verification system evaluate?

The system performs comprehensive checks across five main categories: code correctness, security vulnerabilities, adherence to best practices (like SOLID), performance impact, and documentation quality.

How does the automatic rollback feature work?

If a verification check fails to meet the defined accuracy threshold (default 0.95), the system utilizes Git integration to instantly revert the codebase to the last known good state.

Is the quality threshold customizable?

Yes, you can adjust the verification strictness globally in the config file or on a per-command basis using the --threshold flag (e.g., set to 0.99 for production-critical tasks).

Verification & Quality Assurance

Name: Verification & Quality Assurance
Author: ricable

byricable

•

セキュリティとテスト

Ensures codebase reliability through real-time truth scoring, automated code verification, and an intelligent rollback system.

This skill provides a robust safety layer for AI-driven development by implementing a quantitative 'truth scoring' system and automated quality guardrails. It evaluates agent outputs and code changes against a 0.95 accuracy threshold, performing deep checks for security, performance, and best practices. With integrated features like automatic Git-based rollbacks for failed validations and real-time monitoring dashboards, it enables developers to maintain high-confidence production environments while leveraging AI agents for complex tasks.

主な機能

01Automated security, logic, and best practice verification checks

02Interactive web-based dashboard for quality metrics and trends

03Smart automatic rollback system for changes failing quality thresholds

04Real-time Truth Scoring (0.0-1.0 scale) for code and agent tasks

052 GitHub stars

06Comprehensive CI/CD integration with JSON and HTML reporting

ユースケース

01Validating AI-generated code quality before merging into production

02Monitoring long-term agent performance and reliability trends

03Automating safety gates in continuous integration pipelines

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ricable/ultimate-ai-agent verification-quality

For use in Claude.ai and ChatGPT

Download Skill