How do I install the Debate skill?

Run npx skillmds@latest add nalyk/debate in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Debate safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Debate?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Debate free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Debate?

nalyk (@nalyk) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Debate by nalyk — Agent Skill

Name: Debate
Rating: 4.5 (4 reviews)
Author: nalyk

Debate

Multi-model adversarial debate: Claude defends a position against external CLI models (agy/Gemini, Codex, Mistral Vibe). Trigger on "debate this", "challenge my thinking", "devil's advocate", "stress test this idea", "second opinion from other models", "what am I missing".

nalyk Updated Sep 10, 2026

File contents

Debate System

Claude forms a position, then external AI models challenge it through structured adversarial debate. NOT Claude debating itself — different models = different blind spots = genuine value. Refuses to run with zero external challengers.

Commands

Command	Purpose
`/debate <topic>`	Full adversarial debate
`/debate:doctor`	Probe challenger CLIs, cache the working list
`/debate:adr <decision>`	Debate + formal ADR output

Challengers

CLI	Model	Persona file (in debate workspace)
`agy`	Gemini (Antigravity CLI, pinned to a Gemini model)	`agy/GEMINI.md`
`codex`	GPT (ChatGPT Plus)	`AGENTS.md`
`vibe`	Mistral Vibe	`vibe/AGENTS.md`

The legacy gemini CLI is dead; a resolving binary is a stale shim. Use agy.

Protocol (hybrid)

Workspace + personas → Claude's position → parallel challenge (all CLIs at once via ${CLAUDE_PLUGIN_ROOT}/scripts/challenge-all.sh) → consensus check → sequential confrontation if disagreement → iterate to max rounds → consensus OR tradeoff document with assumption extraction. Disagreement is signal, not failure — the tradeoff output maps which assumption makes each option correct.

Details on demand:

Orchestration: ${CLAUDE_PLUGIN_ROOT}/references/debate-protocol.md
CLI flags/invocation: ${CLAUDE_PLUGIN_ROOT}/references/cli-invocation.md

Settings

~/.claude/debate.local.md YAML frontmatter: max_rounds (default 5), timeout_per_cli (default 120), adr_path (default ./docs/decisions).

Frequently asked questions