A multi-perspective AI debate engine that runs five simulated agent personas in a single LLM call to eliminate confirmation bias.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install multi-perspective-forum
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 multi-perspective-forum 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 Agent Forum?
Agent Forum is a sophisticated decision-making framework built for the Openclaw Skills ecosystem. It facilitates a structured, three-round debate between five distinct AI personas—Revenue Realist, Builder, Skeptic, Operator, and Customer Voice—all within a single LLM execution. This methodology effectively dismantles confirmation bias by forcing diverse viewpoints to clash, providing a level of critical analysis that a single-prompt interaction cannot achieve.
Designed to integrate seamlessly into a developer's workflow, this skill transforms raw data into actionable intelligence. Whether you are analyzing a weekly sprint or evaluating a major pivot, Agent Forum provides a brutally honest assessment of the situation, culminating in a consensus-based verdict and a controllable items assessment to help you focus on what actually moves the needle.
Agent Forum Use Cases
- Execute a Monday Leverage Audit to analyze revenue signals and lead pipelines.
- Conduct a Friday Weekly Retro to compare sprint targets against actual performance.
- Stress-test strategic decisions where confirmation bias poses a significant risk.
- Evaluate new business opportunities identified in logs or external documents.
- Provide a second opinion on opportunity framing when delegation skills require more depth.
How Agent Forum Works
- Gathers data payloads including analytics, lead pipeline counts, and system error logs.
- Constructs a prompt injecting the data and the specific strategic question into the Openclaw Skills engine.
- Initiates a single sessions_spawn call using a high-reasoning model like Claude Opus.
- Simulates a three-round debate where five distinct agent personas challenge each other's assumptions.
- Produces a structured final output containing consensus, contested points, and a controllable assessment verdict.
Agent Forum Setup
The Agent Forum skill is deployed via the Openclaw Skills framework. To configure automated triggers, use the internal OpenClaw cron system rather than standard crontab.
# Example of triggering the forum manually via CLI
openclaw run agent-forum --question "Should we pivot to the new API structure?"
To schedule audits, configure your sessions_spawn with the following cron entries:
- Monday Audit:
0 13 * * 1(9 AM ET) - Friday Retro:
0 13 * * 5(9 AM ET)
Agent Forum Data Schema & Taxonomy
The skill operates on a structured data payload and outputs a specific markdown report based on the following taxonomy:
| Component | Description |
|---|---|
| Data Payload | Input containing analytics, leads, revenue events, and logs. |
| Round 1 | Initial positions from all 5 agents (Revenue, Builder, Skeptic, etc.). |
| Round 2 | Rebuttals and challenges between the conflicting agent perspectives. |
| Round 3 | Final Verdict including Consensus and Contested items. |
| Controllables | A status table using ✅ Controllable, ⚠️ Partial, and ❌ Uncontrollable markers. |
Agent Forum Advanced Features
- Multi-agent simulation within a single sessions_spawn call to maximize token efficiency (~8-12k tokens).
- Automated data collection hooks for standardized Monday Leverage Audits and Friday Retrospectives.
- Direct integration with OUTSTANDING.md for evaluating high-priority business opportunities.
- Hardcoded adversarial perspectives to guarantee a rigorous null-hypothesis check for every decision.
- Native OpenClaw cron support for reliable, scheduled strategic reviews without external dependencies.
SKILL.md
Loading
Related Openclaw Skills
Vybes.fun Solana Launchpad
A comprehensive Solana token launchpad and prediction market platform designed for programmatic AI agent interaction.
Literature Helper
A specialized academic assistant for automated bilingual literature retrieval and deep technical analysis of research papers.
Ghibli Style Travel Map Generator
An AI-powered tool that transforms travel itineraries into hand-drawn Studio Ghibli-style maps with custom POI icons and route visualization.
Recruiter Video
A specialized video assistant for recruiters to demonstrate sourcing expertise and employer partnership value through professional video scripts.
Delegation
A strategic decomposition skill that breaks complex projects into machine-verifiable atomic tasks for multi-model execution and cost optimization.
Seko Video Creation
A professional AI-driven filmmaking tool that automates the entire production workflow from scriptwriting to final video generation.
Featured*
