What is 'ultrathink' in this context?

Ultrathink refers to a high-intensity reasoning mode used by the thinker agent to explore the nuances, potential blind spots, and foundational principles of a given question.

What is the Socratic Self-Refiner skill?

It is a tool that uses a structured questioning and answering loop to deeply analyze and refine complex ideas or problems using specialized agent roles.

How many rounds of refinement does it perform?

The skill typically achieves high-quality results in 2-5 rounds, though it can perform up to 10 rounds if the logic requires further refinement or stagnation is not detected.

Can I use this skill within other automated tasks?

Yes, it is designed with an 'Invoked Mode' that allows other agents to call the Socratic orchestrator to get a refined JSON output for use in broader workflows.

What criteria are used to evaluate the answers?

The skill evaluates answers based on five dimensions: Depth (root principles), Coherence (logical consistency), Completeness (coverage), Practicality (utility), and Humility (limitations).

Socratic Self-Refiner

Name: Socratic Self-Refiner
Author: chkim-su

bychkim-su

0•

生产力与工作流

Refines complex problems and ideas through iterative Socratic questioning and deep analytical reasoning.

Socratic Self-Refiner is a specialized Claude Code skill designed for deep conceptual analysis and high-level problem-solving. It utilizes a dual-agent architecture—an orchestrator that manages a loop between a 'Thinker' for deep analysis and a 'Questioner' for critical evaluation. By subjecting ideas to a multi-dimensional assessment covering depth, logic, and practicality, it systematically polishes initial hypotheses into refined, high-confidence insights. This skill is ideal for architects, researchers, and developers who need to move beyond surface-level answers to uncover root causes and robust, well-vetted solutions.

主要功能

01Structured reporting featuring core insights, limitations, and future exploration paths

02Multi-dimensional quality evaluation across five key analytical metrics

030 GitHub stars

04Deep 'ultrathink' reasoning mode for uncovering root principles

05Iterative Q&A refinement loop with up to 10 rounds of analysis

06Contextual persistence to maintain logic throughout the refinement process

使用场景

01Refining project requirements and problem statements through critical inquiry

02Validating high-level logic and identifying hidden edge cases in complex systems

03Analyzing complex software architectural decisions or ethical dilemmas

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add chkim-su/philosopher socratic-refine

For use in Claude.ai and ChatGPT

Download Skill