Structure multi-round, multi-agent debates to surface tradeoffs and decision rationale.
npx skills add https://github.com/inho-team/qe-framework --skill qdebate
Please help me install this Agent Skill. Skill: Qdebate Source: https://github.com/inho-team/qe-framework/tree/main/skills/Qdebate Command: npx skills add https://github.com/inho-team/qe-framework --skill qdebate
Structured debate on topics using multiple AI agents to surface tradeoffs, rationales, and robust perspectives for decision making.
Provide a topic and optional mode and rounds to initiate a structured multi-agent debate.
High-intent search queries and answers about installing and using this skill.
To run a multi-agent debate, provide a topic, select a mode like agent-vs-agent or codex-vs-claude, and configure the rounds. The system orchestrates staged arguments across independent sides and synthesizes a final summary highlighting the strongest tradeoffs.
Structured multi-agent debate is a process where multiple AI agents argue independent sides of a topic across multiple rounds. Use it for product, engineering, or policy decisions to surface robust rationales and tradeoffs that a single perspective might miss.
Yes, you can route different models for opposing sides using the codex-vs-claude debate mode. This allows you to leverage distinct model perspectives to evaluate design tradeoffs or policy choices.
The best way to get an ADR-ready decision context is to run a configured multi-round debate. The system automatically synthesizes the arguments into a final summary, extracting the strongest tradeoffs and rationale needed for architecture decision records.
You can configure the number of rounds and assign sides by providing optional parameters when initiating the debate. In self-debate mode, a single agent argues against itself across multiple rounds to stress-test a specific position.
No external dependencies or components are required to run structured multi-agent debates. The system handles agent orchestration, side assignment, round management, and final synthesis internally.
louisgilliesAssemble 5-7 historical thinkers for structured debates with filterable attributes.
grasbergDocument architectural decisions using ADR templates and trade-off analysis.
ils15Orchestrate multi-agent system tasks with model routing and sprint management.
lv416eDecompose topics into researchable components and synthesize evidence-based conclusions.
