Which AI models does this skill use?

The skill utilizes Claude Sonnet for initial planning and implementation, and uses Codex as a secondary reviewer to provide specialized feedback on code quality.

Can I customize the coding standards used for reviews?

Yes, the skill references a standards.md file within its reference directory where you can define the specific coding criteria and review rules for your project.

What is the Multi-AI pipeline?

It is a structured workflow that coordinates different AI models to handle specific stages of development: planning, implementation, and specialized code review.

What happens if a review is rejected?

If a reviewer (Sonnet or Codex) identifies issues, the workflow pauses, requiring the AI to fix the issues in the code or plan before restarting the review cycle for final approval.

How does the skill track the progress of a task?

Progress is tracked using a local '.task/' directory containing JSON state files that record the original request, the refined plan, and review outcomes.

Multi-AI Development Pipeline

Name: Multi-AI Development Pipeline
Author: cskwork

bycskwork

開発者ツール

Orchestrates a rigorous development lifecycle that guides AI through planning, multi-stage reviews, and implementation to ensure production-grade code.

概要

The Multi-AI Pipeline skill transforms Claude into a structured project orchestrator, enforcing a 'plan-review-implement-review' workflow for high-reliability software development. By leveraging Claude Sonnet for logic and implementation alongside Codex for specialized review, it ensures every feature is thoroughly architected and validated against coding standards. The skill maintains a persistent state through structured JSON files, requiring sequential approvals from multiple AI agents to minimize errors and maximize architectural consistency in complex codebases.

主な機能

State-driven workflow management using a dedicated .task directory
Structured plan refinement including technical approach and complexity analysis
Automated multi-stage AI review cycle using Sonnet and Codex agents
Strict enforcement of coding standards via reference documentation
Integrated implementation phase with mandatory unit test generation
0 GitHub stars

ユースケース

Architecting complex features that require deep technical planning and double-validation
Managing large-scale refactors where automated multi-agent review reduces regression risks
Enforcing consistent coding standards and best practices across an AI-driven project

概要

主な機能

State-driven workflow management using a dedicated .task directory
Structured plan refinement including technical approach and complexity analysis
Automated multi-stage AI review cycle using Sonnet and Codex agents
Strict enforcement of coding standards via reference documentation
Integrated implementation phase with mandatory unit test generation
0 GitHub stars

ユースケース

Architecting complex features that require deep technical planning and double-validation
Managing large-scale refactors where automated multi-agent review reduces regression risks
Enforcing consistent coding standards and best practices across an AI-driven project