A Mixture of Agents workflow that simulates a rigorous Oxford Union-style formal debate to stress-test arguments and explore complex topics.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install mixtureofagents-debate
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install mixtureofagents-debate using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is moa-debate?
The moa-debate skill implements a sophisticated Mixture of Agents architecture to conduct formal, structured debates within the Openclaw Skills ecosystem. By assigning distinct personas—Proposition, Opposition, Devil's Advocate, and Chair—this skill provides a comprehensive exploration of any given motion. It utilizes specific temperature settings for each role to ensure that main debaters remain rigorous and principled while the Devil's Advocate introduces lateral, adversarial challenges to expose logical weaknesses.
Integrating this into your workflow allows for high-fidelity simulation of parliamentary-style reasoning. The process is far more than a simple back-and-forth; it includes Points of Information (POIs), procedural summaries, and a final completeness judge that ensures all facets of an argument are thoroughly examined. This makes it an essential tool for those using Openclaw Skills to refine their logic or prepare for real-world adversarial environments.
moa-debate Use Cases
- Stress-testing corporate strategies or product directions through adversarial multi-agent reasoning.
- Preparing for formal public speaking engagements or competitive debating by simulating opposition.
- Exploring ethical dilemmas in technology and society using the structured framework of Openclaw Skills.
- Generating comprehensive "for and against" briefs for complex policy decisions or technical architectures.
- Evaluating the robustness of a specific philosophical thesis against a high-temperature Devil's Advocate.
How moa-debate Works
- The user provides a motion in the "This House Believes That..." format and sets optional parameters like completion thresholds and round caps.
- The Proposition and Opposition agents deliver structured opening speeches with evidence-based arguments at a low temperature (0.3) for maximum rigor.
- Agents exchange Points of Information (POIs), with a randomized 60/40 acceptance rate to simulate the unpredictability of live debate.
- A high-temperature Devil's Advocate agent identifies and attacks the current dominant side to prevent one-sided bias and surface hidden assumptions.
- A neutral Chair summarizes each round, while a Completeness Judge evaluates if the strongest arguments on both sides have been adequately addressed.
- Once the completeness threshold is met or the hard round cap is reached, agents deliver closing rebuttals followed by a final structured brief.
moa-debate Setup
To enable moa-debate within your Openclaw Skills environment, ensure your agent has access to the skill definition. No external dependencies are required beyond a standard LLM provider as the skill uses system prompt engineering to differentiate agent roles.
# Example of initiating a debate session via CLI
openclaw run moa-debate --motion "This House Believes That AI should be open-sourced"
You can configure the rigor of the session by adjusting the Min Score to Pass (5-10) and Hard Round Cap (3-8) parameters during the initialization step.
moa-debate Data Schema & Taxonomy
The skill generates structured JSON data for both the judging process and the final output brief. This ensures that Openclaw Skills can store and reference the debate logic programmatically for downstream tasks.
| Component | Format | Description |
|---|---|---|
| Completeness Score | JSON | Contains a 0-10 score and qualitative reasoning for round termination. |
| Debate Brief | JSON | Organized into Pro, Con, Rebuttals, and Floor Attacks. |
| Session Logs | Markdown | A chronological record of all speeches, POIs, and chair summaries for human review. |
| Verdict | Text | A neutral summary from the Chair on the rhetorical and logical weight of each side. |
moa-debate Advanced Features
- Dynamic Temperature Scaling: Assigns specific LLM parameters (0.2 to 0.7) for each role to balance logical consistency with creative challenge.
- Probabilistic Interaction: Features a randomized POI acceptance logic to mirror the human element of parliamentary debate.
- Multi-Agent Synthesis: Uses a Mixture of Agents approach to aggregate the strongest points from multiple perspectives into a single brief.
- Automated Convergence Logic: Employs a dedicated judge agent to determine if the debate has reached logical saturation before the round cap.
- Floor Attack Simulation: Includes a Devil's Advocate specifically designed to identify and target the weakest claim of the dominant side.
SKILL.md
Loading
Related Openclaw Skills
Agent Passport
Agent Passport is a comprehensive security framework providing OAuth-style consent-gating, behavioral contracts, and threat-shielding for autonomous AI agents.
Zongjie: Structured Memory & Experience Management
A structured memory management system that distills complex troubleshooting and development experiences into a permanent knowledge base for AI agents.
Memory Sync Protocol
A standardized workflow for synchronizing agent behavior, routing rules, and long-term preferences across multiple markdown-based configuration files.
Memory Audit Guardian
A governance and optimization utility designed to maintain high-performance memory structures and reduce token bloat in AI agents.
YouTube-Hook
An AI-driven YouTube content strategist that transforms any topic into high-performing video scripts, SEO-optimized metadata, and visual concepts.
Memory Manager
A professional memory architecture for AI agents implementing episodic, semantic, and procedural storage to prevent context loss.
Featured*
