Can this help with debugging subtle bugs?

Absolutely. It forces the AI to consider 'Why might this fix fail?' and 'What am I missing?' which often surfaces the root cause of intermittent or complex issues.

Does this skill change the format of Claude's responses?

Yes, it structures responses into numbered steps with explicit sub-sections for self-critique, confidence levels, and remaining uncertainties.

When should I use the Self-Doubt skill?

Use it for complex problems where errors are costly, such as security-sensitive code, complex refactoring, or when you need to double-check high-level logic.

How does the Recovery Protocol work?

If Claude realizes a previous step in its reasoning was flawed, the protocol dictates that it must stop, acknowledge the error, and restart the logic from the corrected point.

Does this make the AI more hesitant?

It makes the AI more honest about its uncertainty, preferring to state 'I'm not certain' over providing a confident but incorrect answer.

Critical Reasoning & Self-Doubt

Name: Critical Reasoning & Self-Doubt
Author: oxhagolli

byoxhagolli

•

生産性とワークフロー

Enhances Claude's analytical rigor by applying a self-critical reasoning protocol to identify logical flaws and hidden assumptions.

The Self-Doubt skill transforms Claude into a meticulous problem-solver by enforcing a structured, skeptical reasoning protocol. It requires the AI to explicitly question every step of its logic, assign confidence levels to its claims, and systematically identify potential edge cases or assumptions that could lead to errors. This skill is indispensable for complex technical architecture, high-stakes decision-making, or debugging subtle logical bugs where overconfidence typically leads to costly mistakes, ensuring every output is battle-tested through internal critique.

主な機能

01Quantified confidence levels for individual steps to highlight uncertainty

02Automated recovery protocol to stop and re-examine work when errors are detected

031 GitHub stars

04Step-by-step reasoning with mandatory self-critique for every logical jump

05Systematic evaluation of factual, logical, and assumption-based doubts

06Detailed uncertainty reporting including suggested verification methods

ユースケース

01Double-checking mission-critical infrastructure changes before deployment

02De-risking architectural decisions by exploring potential failure modes and edge cases

03Verifying complex mathematical or logical proofs for software algorithms

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add oxhagolli/clawdskillz self-doubt

For use in Claude.ai and ChatGPT

Download Skill