Summary
This walks you through five layers of questioning to examine beliefs, decisions, or proposals without lecturing. It clarifies terms, surfaces hidden assumptions, probes evidence, explores implications, and considers counterarguments. The output format is structured with tables and reflection points at each layer, ending with a refined position and key insights. Useful when you need to stress test an idea before committing to it, or when you're teaching someone and want them to arrive at understanding themselves. The steel man section is a nice touch. It takes patience to work through properly, but that's kind of the point.
Install to Claude Code
npx -y skills add neurofoo/agent-skills --skill socratic --agent claude-codeInstalls into .claude/skills of the current project.
Plug Mailtrap into your AI workflow and let it handle the email.
Connect Mailtrap MCP →
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Make coding agent sessions - Searchable, Shareable, Vendor-neutral & Scored.
Try For Free →
Your agent targets a perfect 10 Code Health score. Deterministic. Every commit.
Try For Free →
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
Files
SKILL.mdView on GitHub
Socratic Questioning
Guide discovery through systematic questioning, helping examine a belief, decision, or idea more deeply.
Instructions
Work through multiple levels of questioning to explore the topic. The goal isn't to prove the person wrong—it's to help them (or yourself) think more clearly and deeply.
Output Format
Topic/Belief: [What we're examining]
Starting Point
The Claim/Position [Restate the belief or idea being examined]
Layer 1: Clarifying the Claim
Questions
- What exactly do you mean by [key term]?
- Can you give a specific example of this?
- What would this look like in practice?
Reflections [What these questions reveal about the claim]
Layer 2: Probing Assumptions
Hidden Assumptions
| Assumption | Question to Test It |
|---|---|
| [assumption 1] | "Is it always true that...?" |
| [assumption 2] | "What would need to be true for...?" |
Questions
- What are you taking for granted here?
- Why do you think [X] leads to [Y]?
- What if [assumption] weren't true?
Layer 3: Examining Evidence
Questions
- What evidence supports this view?
- How do you know that [claim]?
- What would change your mind?
- Is there any evidence against this?
Layer 4: Exploring Implications
Questions
- If this is true, what else follows?
- What are the consequences of believing this?
- How does this apply to [edge case]?
- If we're wrong about this, what's the cost?
Layer 5: Alternative Perspectives
Questions
- What would someone who disagrees say?
- How would [specific person/role] view this?
- What's the strongest argument against?
Steel Man The best argument against this position:
[Strongest counterargument]
Synthesis
Refined Position After questioning, a more nuanced view might be:
[Refined statement]
Remaining Questions
- [Question 1]
- [Question 2]
Key Insight The most valuable thing this questioning revealed:
[Insight]
Guidelines
- Ask, don't tell
- Follow curiosity, not agenda
- "I don't know" is a valuable answer
- The goal is understanding, not winning
- Be patient—good questions take time to answer
$ARGUMENTS
Featured
Plug Mailtrap into your AI workflow and let it handle the email.
Connect Mailtrap MCP →Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →Make coding agent sessions - Searchable, Shareable, Vendor-neutral & Scored.
Try For Free →Your agent targets a perfect 10 Code Health score. Deterministic. Every commit.
Try For Free →Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
