What is the Audit-and-Fix mode?

In this mode, the skill not only identifies quality issues but also proceeds to fix them directly, following a strict RED->GREEN protocol to prevent regressions.

What is a six-dimension review in task-ai?

It is a comprehensive evaluation framework that assesses Correctness (D1), Security (D2), Reliability (D3), Performance (D4), Architecture (D5), and Maintainability (D6).

Can this skill be used outside of a formal task lifecycle?

Yes, it includes a 'scope=context' mode that allows you to review any current conversation plan, code snippet, or solution approach on demand.

What are anchored reviews?

Anchored reviews compare deliverables against specific project files like .target.md (requirements) and .plan.md (design) to ensure objective, non-subjective scoring.

How does gated execution work?

The skill processes quality gates sequentially; if a critical gate like Security or Correctness fails, the process is blocked until a fix is implemented and verified.

Gated Quality Audit

Name: Gated Quality Audit
Author: huacheng

byhuacheng

•

Developer Tools

Performs rigorous six-dimension gated reviews to evaluate plans, implementations, and code quality against anchored requirements.

The task-ai:check skill is a sophisticated quality assurance tool designed for AI-driven development lifecycles. It employs a Six-Dimension (D1-D6) gated review process—covering Correctness, Security, Reliability, Performance, Architecture, and Maintainability—to ensure that every phase of a task meets specific quality thresholds. By anchoring evaluations to project-specific requirements and convergence baselines, it provides objective scoring and identifies blocking issues. Its unique 'Audit-and-Fix' mode even allows the AI to automatically remediate findings using a mandatory RED->GREEN regression testing protocol, ensuring high-integrity deliverables.

Key Features

01Context-aware evaluation for technical plans, bug fixes, and architectural designs

02Automated 'Audit-and-Fix' mode with mandatory RED->GREEN regression testing protocol

03Six-dimension (D1-D6) quality gates covering Security, Correctness, and Reliability

04Four-file anchored review mapping deliverables against requirements, plans, and baselines

051 GitHub stars

06Convergence tracking with numeric scoring (0.0 - 1.0) and threshold enforcement

Use Cases

01Performing mid-execution assessments to ensure implementation aligns with technical plans

02Automating the validation of custom Claude Code skills and evolving rule candidates

03Conducting a pre-implementation security and feasibility audit for complex features

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add huacheng/moonview check

For use in Claude.ai and ChatGPT

Download Skill