How does the review process work?

The skill executes sequential reviews. It first invokes Sonnet for a review, and once passed, invokes Codex. Any issues found must be fixed before the pipeline can proceed to implementation.

Where are the task files and states stored?

All pipeline data, including the user request, technical plans, and implementation results, are stored in a local `.task/` directory within your project folder.

Can I skip the planning phase?

No, the skill follows a strict protocol that requires a refined plan and successful reviews before any code implementation begins.

What is the Multi-AI pipeline?

It is a structured workflow that coordinates multiple AI models to plan, refine, implement, and review code changes systematically to ensure high reliability.

Which AI models are used in this skill?

This skill primarily uses Claude (Sonnet) for initial planning and coding, and Codex as a secondary reviewer to provide specialized feedback.

Multi-AI Workflow Pipeline

Name: Multi-AI Workflow Pipeline
Author: cskwork

bycskwork

0•

Productivity & Workflow

Automates a structured multi-AI pipeline for planning, implementing, and reviewing code using Claude and Codex models.

The Multi-AI Pipeline skill transforms the development process into a rigorous, multi-stage workflow designed to ensure high-quality code output. It coordinates between Claude (acting as the planner and coder) and Codex (acting as the specialized reviewer) to manage tasks from initial requirement gathering through final implementation. By enforcing a strict process—including plan refinement, mandatory cross-model code reviews, and state-managed transitions—it minimizes logic errors and ensures that every change aligns with the project's technical architecture before the final commit.

Key Features

010 GitHub stars

02Automated state management and task cleanup

03Mandatory dual-AI review process using Sonnet and Codex

04Integrated implementation and automated testing workflow

05Robust JSON-based task tracking and state persistence

06Structured multi-stage planning and requirement refinement

Use Cases

01Executing complex feature implementations requiring rigorous technical planning

02Automating the transition from raw requirements to production-ready code

03Enforcing high-standard code reviews using specialized reviewer models like Codex

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add cskwork/llm-review multi-ai

For use in Claude.ai and ChatGPT

Download Skill