Do I need an OpenAI Codex API key to use this skill?

Codex is recommended for the most objective 'Red Team' experience, but the skill includes an automatic 'Claude-only' fallback mode that works without external dependencies.

How does Devil's Advocate differ from standard code review?

While code review focuses on implementation, Devil's Advocate focuses on the underlying design and logic before implementation starts, using a formal adversarial debate to find conceptual flaws.

When should I use this instead of the Strong Inference skill?

Use Strong Inference for investigating unknown bugs or finding root causes; use Devil's Advocate for validating new proposals, architectural designs, and strategic decisions.

Can I customize the intensity of the review?

Yes, you can use the --max-rounds flag to control the depth of the debate and the number of iterative improvements required before a final verdict is reached.

Devil's Advocate

Name: Devil's Advocate
Author: masuP9

bymasuP9

•

Productivity & Workflow

Stress-tests technical designs and proposals through structured adversarial debate to identify risks and refine decision-making.

The Devil's Advocate skill is a powerful validation tool that subjects your technical proposals, architectural designs, and hypotheses to a systematic 'Red Team vs. Blue Team' review process. By simulating critical opposition—optionally leveraging OpenAI Codex as an independent critic—it exposes hidden assumptions, technical risks, and edge cases before implementation begins. This structured approach ensures higher-quality engineering decisions, more robust systems, and a clear audit trail of the reasoning behind your most critical design choices.

Key Features

01Collaborative mode using OpenAI Codex as an independent external critic

02Multi-round refinement process to address and mitigate identified risks

03Severity-based categorization of concerns from Critical to Low

042 GitHub stars

05Persisted debate state with formal final verdicts (Approve, Conditional, Reject)

06Structured Red Team vs. Blue Team adversarial debate methodology

Use Cases

01Conducting objective risk assessments for complex technical migrations

02Validating high-level system architecture and infrastructure changes

03Stress-testing API designs or security proposals before coding begins

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add masup9/codex-collab devils-advocate

For use in Claude.ai and ChatGPT

Key Features

01Collaborative mode using OpenAI Codex as an independent external critic

02Multi-round refinement process to address and mitigate identified risks

03Severity-based categorization of concerns from Critical to Low

042 GitHub stars

05Persisted debate state with formal final verdicts (Approve, Conditional, Reject)

06Structured Red Team vs. Blue Team adversarial debate methodology

Use Cases

01Conducting objective risk assessments for complex technical migrations

02Validating high-level system architecture and infrastructure changes

03Stress-testing API designs or security proposals before coding begins

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add masup9/codex-collab devils-advocate

For use in Claude.ai and ChatGPT