Can I customize the perspectives involved in the debate?

Yes, you can configure the number and types of perspectives, ranging from a default set (Security, Performance, Simplicity) to comprehensive sets including Scalability and Cost.

What happens after a decision is made?

The skill provides implementation guidance and can use builder agents to begin executing the decided path, including setting up monitoring for success metrics.

Does this skill help with documentation?

Absolutely. It automatically generates a detailed decision record in your repository, including the full debate transcript, rationale, and implementation guidance.

How does the skill handle conflicting viewpoints?

It uses configurable convergence criteria such as synthesis, majority rule, or evidence-based ranking to reach a final, justified recommendation.

When should I use the Debate Workflow skill?

Use it for high-impact decisions like framework selection, major refactoring, or infrastructure changes where multiple valid approaches exist and require deep evaluation.

Architectural Debate Workflow

Name: Architectural Debate Workflow
Author: rysweet

byrysweet

•

Productivity & Workflow

Facilitates structured multi-perspective debates among AI agents to resolve complex architectural trade-offs and technical decisions.

The Debate Workflow skill enables Claude to conduct rigorous, multi-agent analysis for high-stakes engineering decisions. By simulating various expert perspectives—such as Security, Performance, and Cost—this skill breaks down complex problems into structured rounds of positioning, challenging, and synthesis. It is particularly useful for framework selection, major refactors, and system design choices where a single 'correct' answer isn't immediately obvious, ensuring all trade-offs are documented and evaluated before implementation to prevent long-term technical debt.

Key Features

01Simulates up to 7 expert agent perspectives including Security, Scalability, and Cost

02Automatic generation of standardized Decision Records (ADRs) with full transcripts

03Evidence-based argument tracking using specialized analyzer and architect agents

04Multi-agent subprocess coordination to ensure unbiased and independent viewpoints

05Configurable debate rounds ranging from quick positioning to deep synthesis cycles

0616 GitHub stars

Use Cases

01Documenting high-stakes infrastructure decisions for long-term stakeholder alignment

02Resolving team disputes over testing strategies and technical debt management

03Evaluating major architectural shifts like REST vs. GraphQL or Monolith vs. Microservices

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rysweet/amplihack debate-workflow

For use in Claude.ai and ChatGPT

Download Skill