Socratic AI agent prompting skill

socratic-ai-prompt-skill

Socratic prompting is a strategy that shifts AI from a "vending machine" for answers to a "thinking partner". Instead of providing a direct answer, the AI is instructed to ask questions, challenge assumptions, or lead you to a conclusion through dialogue. It follows the socratic method principles.

Core Principles

• Interrogation Before Execution: Force the AI to ask clarifying questions about your goal before it attempts to solve it.
• Assumption Surfacing: The AI identifies hidden beliefs or unstated constraints in your request.
• Iterative Refinement: You and the AI engage in a back-and-forth "dance" to narrow down the best possible solution.
• Role Reversal: You can act as the student while the AI acts as a Socratic Tutor that guides you to the answer rather than giving it to you.

How To Use It

This skill intercepts a standard request, performs domain research (if necessary), transforms the request into a rigorous Socratic dialogue framework, and requires your sign-off before execution. <br>

Example:

Once the skill is added to your Agent configuration:

1. Trigger it:

"Use the Socratic Architect to help me design a scalable microservices architecture."

2. Observe Research: Agent will likely run a search command like: google_search("common pitfalls in microservices scaling").
3. Review the Draft: Agent will present a prompt: "Act as a Senior Architect. Challenge my choice of database consistency models. Ask me to justify why a monolith wouldn't suffice..."
4. Prompt refinement loop:

• added constraint - we use PostgreSQL
• I have refined the Socratic inquiry. Should I execute this to begin our dialogue or would you like to further adjust the technical focus?

5. Approve & Execute: "Execute!" <br> Agent immediately switches modes: `"Understood. Before we discuss scaling, what specific failure domains are you trying to isolate by splitting you