What does 'wiring verification' mean in this skill?

Wiring verification is the process of tracing code paths to confirm that new features or functions are actually connected to the appropriate entry points and are being invoked correctly by the existing system.

Can this skill be customized with more agents?

Yes. While three agents are the minimum to ensure diversity of thought, you can include additional specialized agents for areas like security, performance, or specific framework best practices if they are available.

Why use multiple agents instead of just one for a code review?

Using different models and specialized roles (such as Architect and Codex agents) provides broader coverage and diverse perspectives, catching subtle logic and structural issues that a single-agent review might overlook.

Should I run this before or after my unit tests?

This skill is intended to be used after your implementation is complete and your local linting and tests have already passed, serving as a final high-level logic and architectural audit.

Multi-Agent Change Reviewer

Name: Multi-Agent Change Reviewer
Author: hughescr

byhughescr

•

Herramientas de Colaboración

Conducts exhaustive, multi-perspective code reviews of uncommitted changes using parallel AI agents to ensure architectural integrity and bug-free implementation.

The Review-Changes skill automates the critical quality assurance step between code implementation and the final commit. By orchestrating at least three specialized AI agents—including a Codex-based reviewer, a code quality expert, and a structural architect—it verifies that uncommitted code matches the intended plan, identifies dead code, and ensures all new features are correctly wired into the system's call graph. This multi-model approach provides a level of scrutiny that exceeds single-pass reviews, catching subtle logic errors and unintended side effects that might be missed during manual inspection.

Características Principales

01Parallel Multi-Agent Execution: Leverages at least three distinct AI models for diverse perspectives and specialized scrutiny.

021 GitHub stars

03Architectural Wiring Verification: Traces call graphs to ensure new features are properly integrated and actually invoked by the system.

04Contextual Plan Alignment: Compares current uncommitted changes against the original project plan to ensure requirement fulfillment.

05Dead Code Detection: Automatically identifies and flags redundant logic or orphaned code segments for pruning.

06Consolidated Severity Reporting: Summarizes findings into a single report categorized by critical, major, and minor issues.

Casos de Uso

01Final pre-commit quality gate to ensure new code meets architectural and logic standards.

02Automated bug hunting and logic verification after local tests and linting have passed.

03Verifying complex feature integrations where manual call-graph tracing is difficult and error-prone.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hughescr/claude-code-config review-changes

For use in Claude.ai and ChatGPT

Download Skill