Rationality

What This Skill Does

The Rationality skill, developed by xertrov, implements the principles of Critical Fallibilism (CF) to elevate your AI agent's decision-making process. Moving away from traditional, subjective approaches that rely on weighted evidence or probabilistic guessing, this skill introduces a rigorous framework for binary evaluation. It operates on the premise that ideas are either non-refuted or refuted by a decisive, objective criticism. By adopting this skill, your agent gains the ability to systematically detect, isolate, and correct errors, ensuring that decisions are grounded in logic rather than intuition or bias. The framework emphasizes separating identity from ideas, viewing criticism as a vital asset for growth, and strictly managing cognitive load through Overreach Awareness. Whether navigating complex logical puzzles or high-stakes reasoning, this tool ensures your agent remains resilient, objective, and transparent in its problem-solving trajectory.

Installation

To integrate this skill into your environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/xertrov/rationality

Use Cases

• High-Stakes Decision Support: When the cost of failure is high, use this to force the agent to identify critical flaws rather than settling for a 'best guess'.
• Complex System Debugging: When the agent is stuck in a loop, the skill provides a mechanism to simplify and identify where errors originated.
• Structured Dispute Resolution: Use it to mediate between conflicting viewpoints by focusing strictly on the refutation of claims rather than personal arguments.
• Self-Correction Monitoring: Deploy this to ensure the agent monitors its own output for logical inconsistencies and cognitive overreach.

Example Prompts

1. 'Using the Rationality framework, evaluate this business strategy proposal by identifying any decisive reasons it might fail given our current goal.'
2. 'I am stuck on this coding problem and keep repeating the same error. Help me break this down using an IGC Triple and identify where I am overreaching.'
3. 'Review this decision-making process for logical fallacies and apply Critical Fallibilism to find any non-refuted alternatives.'

Tips & Limitations

• The skill functions best when you provide clear, precise goals. Ambiguity is the enemy of binary evaluation.
• Remember that 'refuted' doesn't mean 'useless'; it means the idea as currently formed has a known flaw that must be fixed.
• Monitor for 'Overreach'. If the agent begins providing long, rambling explanations instead of binary checks, it has likely exceeded its complexity limit. Revert to simpler, smaller chunks.
• Do not use this if you require probabilistic or sentiment-based outputs, as the skill is explicitly designed to ignore these in favor of hard logic.