Which AI models does the Agent Reviewer use?

This skill specifically orchestrates parallel reviews using Codex and Gemini to provide a diverse range of technical and editorial feedback.

Can I use this without a Linear account?

Yes, the skill supports both Linear as a task provider and local storage modes using Markdown files in your repository.

Is this skill intended to be used standalone?

While it can be invoked directly, it is primarily designed as a worker in the ln-310 story validation pipeline to handle Phase 5 agent reviews.

How does the 'debate' feature work?

Claude acts as a primary validator, reviewing the output from external agents and challenging them if the suggestions are inconsistent or low-confidence before presenting the final verdict.

Multi-Agent Story Reviewer

Name: Multi-Agent Story Reviewer
Author: levnikolaevich

bylevnikolaevich

•

147

•

Collaboration Tools

Executes parallel external agent reviews using Codex and Gemini to critically verify and improve project stories and tasks.

The Multi-Agent Story Reviewer is a specialized worker skill designed for the ln-310 validation pipeline that orchestrates simultaneous reviews from multiple LLM providers. By running parallel background tasks through Codex and Gemini, the skill performs a cross-model debate to identify technical inconsistencies, security risks, and architectural flaws in your requirements. It filters and deduplicates findings to provide high-confidence editorial suggestions, ensuring your stories and tasks are production-ready before development begins.

Key Features

01Automated health checks and prompt synchronization to ensure consistent review standards

02Real-time background processing that handles agent results as they arrive

03Parallel agent execution using Codex and Gemini for multi-perspective validation

04Seamless integration with both Linear API and local Markdown-based task management

05Internal debate logic where Claude critically verifies and challenges agent suggestions

06147 GitHub stars

Use Cases

01Finalizing high-stakes technical specifications before moving to the implementation phase

02Reducing development rework by identifying requirement gaps through AI-driven debate

03Automating the peer-review process for project stories in a large-scale delivery workflow

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add levnikolaevich/claude-code-skills ln-311-agent-reviewer

For use in Claude.ai and ChatGPT

Download Skill