Council Review
Run any question, plan, PR, or code through a Diverse Multi-Agent Debate (DMAD) council of 5 AI advisors with distinct reasoning methods. Advisors collaborate, peer-review each other anonymously, and a chairman synthesizes a verdict. Empirically outperforms adversarial debate (M3MADBench 2026, DMAD ICLR 2025). Use when: 'council this', 'run the council', 'council review', 'pressure-test this', 'stress-test this', 'war room this', or when facing a genuine decision with stakes and tradeoffs.
Install
npx skills add ngmeyer/council-review --skill council-reviewgit clone --depth 1 https://github.com/ngmeyer/council-review.git /tmp/council-review
mkdir -p ~/.claude/skills
cp -r /tmp/council-review/. ~/.claude/skills/council-reviewPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Council Review
Run any question, plan, or code through 5 independent advisors who use **distinct reasoning methods**, collaborate to refine answers, peer-review each other anonymously, and synthesize a verdict you can trust.
This skill implements the **Diverse Multi-Agent Debate (DMAD)** pattern. It is collaborative, not adversarial: agents seek truth through diversity of reasoning, not by arguing opposing positions.
## Why This Works (Research Backing)
- **Method diversity beats single-method debate.** DMAD (ICLR 2025) shows that agents using distinct reasoning methods reliably outperform homogeneous councils — diverse medium-capacity models can beat GPT-4 on GSM-8K (91% vs 82%) when each agent applies a different reasoning approach.
- **Collaborative debate beats adversarial debate.** M3MADBench (2026) shows that across all modalities, collaborative DMAD outperforms adversarial Div-MAD "by a substantial margin." Adversarial paradigms introduce divergent noise; for **open questions, plans, and decisions**, collaborative deliberation is the right tool.
- **Anonymous peer review prevents provider bias.** Universal across the literature — reviewers defer to role names if visible, so peer-review responses must be shuffled.
- **Confidence calibration breaks the martingale ceiling.** Vanilla MAD often underperforms simple majority vote; confidence-modulated updates ("Demystifying MAD" 2026) systematically drift the council toward correct answers.Details
- Skill name
council-review- Repository
- ngmeyer/council-review
- License
- MIT
- Repo updated
- 31 May 2026
FAQ
What does the Council Review skill do?
Run any question, plan, PR, or code through a Diverse Multi-Agent Debate (DMAD) council of 5 AI advisors with distinct reasoning methods. Advisors collaborate, peer-review each other anonymously, and a chairman synthesizes a verdict. Empirically outperforms adversarial debate (M3MADBench 2026, DMAD ICLR 2025). Use when: 'council this', 'run the council', 'council review', 'pressure-test this', 'stress-test this', 'war room this', or when facing a genuine decision with stakes and tradeoffs.
How do I install the Council Review skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add ngmeyer/council-review --skill council-review. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Council Review popular?
It lives in ngmeyer/council-review, which has 17 GitHub stars and 4 forks.
Related skills
Apple App Review Compliance
20yohannescodes/apple-app-review-compliance
Audit Apple-platform applications, proposed features, code changes, App Store metadata, privacy practices, and submission readiness against Apple's current App Review Guidelines. Use during feature planning, PR review, pre-release QA, App Store rejection triage, or whenever StoreKit, HealthKit, permissions, authentication, user-generated content, or App Review policy matters.
Assumptions
20Teycir/Assumptions
Generate an evidence-backed Assumptions ledger for a Git diff, feature, endpoint, worker, migration, or code path. Use when reviewing a change, asking what could break, preparing a release, finding edge cases, assessing deployment safety, analyzing retries or concurrency, or designing failure-focused tests.
Taiwan Exam Generator
19niansia/taiwan-exam
Create original Taiwan GSAT/CAP (學測／會考) exams for 國綜、國寫、英文、數A、數B、社會、自然, with separate question/solution PDFs and official scope, difficulty, originality, answer, visual and layout QA.
Plan and Build
17jattanjie21/plan-and-build
Use when a user wants to plan and implement a new feature or bug fix from scratch. Triggers on requests like "I want to build X", "add a feature for Y", "fix Z", or when the user invokes /plan-and-build. Guides through clarifying questions, breaks work into small features, sets up a tasks folder, implements each feature with TDD (test first), and produces a manual testing guide when complete.
Gan
15GAN-Thinking/gan-skill
Adversarial dialectic tool inspired by GAN. Alternates between Generator (creative advocate) and Discriminator (adversarial critic). Supports forced role selection (g/d), intensity modes (hard/soft), and multi-language output. Use for stress-testing ideas, business plans, code, and spec designs.
Qikly
19gal-a/qikly
Write tests that can actually fail, by withholding the acceptance criteria from the agent that writes the code. Use when someone does not trust a suite that passes. Use when they want tests written from a specification rather than from the code. Use when they ask whether a specification is testable, or want an existing suite scored by planting faults in the code. Python modules, through the qikly tool. Also use whenever qikly or spec-driven testing is mentioned.
Browse the most popular Claude skills, all testing & qa skills, or write your own with the skills generator.