Can I add more than two teams to the debate?

Yes, you can configure additional teams including Green (Pragmatic/Implementation), Yellow (Risk/Safety), and Purple (Synthesis/Integration) to broaden the scope of the review.

Does this skill modify my source code during the debate?

No, the subagents are explicitly configured with read-only tools. They can research your code and documentation but cannot make modifications.

Where are the debate results stored?

All arguments, rebuttals, and conclusions are saved in a structured Markdown file within a dedicated './debates/' directory in your project.

How do I start a new debate session?

You can initiate the workflow by using the command '/debate {topic}' within your Claude Code environment.

What is a 'Red vs. Blue' debate in this context?

It is a structured exercise where the Red Team acts as the skeptic or challenger of a proposal, while the Blue Team acts as the advocate or defender, ensuring all pros and cons are surfaced.

Debate Orchestrator

Name: Debate Orchestrator
Author: srnnkls

bysrnnkls

0•

Collaboration Tools

Orchestrates structured multi-perspective technical debates using adversarial AI subagents to evaluate architectural trade-offs and technology choices.

The Debate Orchestrator skill enables developers to stress-test ideas by spawning autonomous subagents that represent diverse perspectives, such as Red (skeptic), Blue (advocate), and Green (pragmatic) teams. It automates a rigorous, multi-step adversarial process—including parallel research, evidence-backed argumentation, and sequential rebuttals—all coordinated via a centralized scratchpad. This is an essential tool for high-stakes decision-making, allowing teams to uncover hidden risks, validate architectural assumptions, and reach a documented consensus before writing a single line of production code.

Key Features

01Automated moderator patterns to ensure debate health, depth, and relevance

020 GitHub stars

03Synthesis rounds to identify common ground and actionable recommendations

04Parallel subagent research phases using web search and codebase analysis

05Multi-team configuration including Red, Blue, Green, Yellow, and Purple perspectives

06Structured scratchpad logging in Markdown for persistent decision records

Use Cases

01Comparing third-party libraries or frameworks for specific project requirements

02Evaluating competing architectural patterns (e.g., Microservices vs. Monolith)

03Conducting 'Devil's Advocate' risk analysis for security-sensitive features

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add srnnkls/tropos debate-start

For use in Claude.ai and ChatGPT

Download Skill