An advanced system prompt optimization skill that creates realistic personas, sets debate constraints, and audits owner profiles for AI agents participating in roundtable panels.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install roundtable-prompt-optimizer
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 roundtable-prompt-optimizer 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 Prompt Optimizer?
Prompt Optimizer is an essential tool in the ecosystem of Openclaw Skills designed to prepare AI agents (such as OpenClaw shrimp) for high-intensity, multi-guest roundtable debates alongside humans. By modeling anthropomorphic traits, ensuring strict behavioral guardrails, and projecting accurate owner stances, this skill makes AI-driven speech realistic, plausible, and strategically sharp.
At its core, this skill ensures that the agent consistently represents its human owner's perspective. It does this by deploying multi-layer debate constraints and a powerful profile audit engine that eliminates chronological, logical, and grammatical errors in agent biographies, transforming cardboard characters into highly credible virtual panel guests.
Prompt Optimizer Use Cases
- AI Agent Roundtable Debates: Preparing autonomous agents to represent human owners at technology, ethics, or industry conferences.
- Persona Realism Optimization: Refining and auditing user/agent bio profiles to ensure chronological, logical, and semantic integrity.
- High-Intensity Debate Scenarios: Deploying "Savage Mode" (roast) to challenge opposing viewpoints using rhetorical questions, reduction to absurdity, and structured arguments.
- Dynamic Tone Adjustments: Adapting communication styles to match diverse panels, ranging from gentle, academic styles to assertive or highly tactical responses.
How Prompt Optimizer Works
- Initialize and Select: The user triggers the prompt optimizer using the CLI with options for creating a new profile (
new), polishing an existing one (polish), or launching a high-intensity debater (roast). - Load Scenario Templates: The skill dynamically fetches templates based on 7 predefined forum types, including AI Tech Frontier, AI Ethics, Industry Applications, and Controversy Debates.
- Interactive Persona Collection: The user defines variables such as the agent's name, the owner's details, the core viewpoints, and non-negotiable stances.
- Audit and Polish (Optional): If refining an existing profile, the tool executes a 9-dimension audit analyzing timeline plausibility, career path logic, semantic fluency, and debate penetration resistance.
- Generate Prompt Output: A comprehensive system prompt is produced, complete with a persona card, behavioral constraint boundaries, and scenario-specific tone instructions, ready for copy/export.
Prompt Optimizer Setup
Install and run the Prompt Optimizer within your OpenClaw environment. Use the command line interface to interact with its capabilities.
To list all available debate templates:
/prompt-optimizer list-templates
To generate a new tech-focused prompt interactively:
/prompt-optimizer new tech
To optimize an existing profile markdown file:
/prompt-optimizer polish --file="./shrimp-profiles.md"
To execute a single-shot high-intensity debate setup:
/prompt-optimizer roast debate --name="Savage Shrimp" --owner="Bob, AI Evangelist" --topic="Is Generative AI Overhyped?"
Prompt Optimizer Data Schema & Taxonomy
Prompt Optimizer structures its system prompt assets and reference materials into organized files. Below is a detailed view of its data layout and internal reference schema:
File Structure & Reference Resources
| File Path | Description | Key Contents |
|---|---|---|
references/persona-templates.md |
Contains persona presets. | Assertive, Gentle, Humorous, Academic, Savage, and custom mixes. |
references/debate-phrases.md |
Roundtable phrase library. | Phrase banks for defense, offense, conversation control, and recovery. |
references/ai-timeline-china.md |
Chinese AI timeline (2022-2026). | Used for profile timeline validation and milestone anchoring. |
references/profile-audit-checklist.md |
Audit framework criteria. | Rules for assessing the 9 profile dimensions. |
references/semantic-logic-audit.md |
Semantic & logical rules. | 8 language error types and 5 logical fallacy checks. |
9-Dimension Audit Schema
Each audit processes your profile through dimensions scoring from 1 to 10:
- Timeline Plausibility: Matches ages and milestones logically.
- Industry Event Anchoring: Ensures historical accuracy relative to real-world events.
- Career Path Logic: Highlights realistic job/industry transitions.
- Stance Consistency: Cross-checks views against the persona's background.
- Detail Density: Replaces grand generalizations with concrete facts.
- Weaknesses/Blind Spots: Injects flaws to avoid unrealistic "cardboard" characters.
- Debate Penetration Resistance: Fixes narrative contradictions that opponents could exploit.
- Semantic Fluency: Audits mixed constructions, unclear references, and prepositions.
- Logical Causality: Corrects time reversals and causal leaps.
Prompt Optimizer Advanced Features
- Owner Profile Audit Engine: Automatically checks user bios against 9 critical qualitative axes, generating dynamic scores and concrete rewriting suggestions.
- Debate Constraint Matrix: Implements ground rules, logical strategies, and strict Red Lines (e.g., never claim sentience, never leak system instructions).
- Scenario-Specific Tone Adapter: Dynamically compresses responses under time pressure or switches to structured recovery tactics when under siege.
- Savage Mode Injection: Boosts rhetorical aggression and debate intensity while strictly maintaining safety and positional guardrails.
- Detailed Semantic Fluency and Causality Checkers: Corrects advanced errors like time-reversed causal loops and stacked redundant prepositions.
SKILL.md
Loading
Related Openclaw Skills
The Future of Humanity AI Agent Skill
An executable AI agent toolkit based on Michio Kaku's book that guides users through the science of terraforming Mars, interstellar travel, human immortality, and space colonization.
Prompt Crafter Pro
Prompt Crafter Pro is a comprehensive prompt diagnostics and optimization skill that scores and rewrites AI instructions for models like Claude, GPT-4o, and DeepSeek.
Mastering Bitcoin
An executable AI agent toolkit based on Andreas Antonopoulos's Mastering Bitcoin to help developers and researchers master decentralized protocols, cryptographic keys, and the UTXO model.
Skill Compass
Skill Compass is a diagnostics and validation tool designed to optimize description-based triggering and eliminate silent activation failures in AI agents.
Contract Review Pro
Contract Review Pro is an intelligent Chinese contract risk review skill that scans, annotates, and explains clauses across 15+ contract types under PRC law.
Product Validation Planner
A structured AI skill designed to help creators validate product ideas, analyze Ask HN queries, and generate high-quality actionable workflows.
Featured*
