Argue runs structured debates where multiple AI agents analyze a question independently, cross-examine each other's claims across rounds, and converge on a consensus using voting. It produces detailed reports, event streams, and rendered summaries.
Use for strategic, high-stakes, or ambiguous decisions (architecture trade-offs, risk analysis, major product choices) where multiple perspectives and cross-examination yield stronger conclusions. Not suitable for quick factual lookups or time-critical tasks.
Designed for CLI-based agent providers and multi-model setups (Codex/Copilot adapters, Gemini CLI, Claude CLI, or other provider adapters).
Argue is a multi-agent debate engine CLI that orchestrates structured debates across AI agents with cross-examination rounds and consensus voting. The SKILL.md is well-structured with clear triggers, pre-flight setup, usage examples, flag reference, output file documentation, and troubleshooting. No scripts were included — the skill wraps the argue CLI entirely through command invocation. The global npm install is the only security-adjacent concern, but it explicitly instructs user confirmation first.
Clean skill, no scripts to audit. CLI-wrapper pattern done well. Good progressive disclosure with references/ directory for advanced setup. No security concerns beyond the standard npm global install.