Agent Skills: Socratic Questioning

Guide discovery through questioning techniques and pattern recognition for Clean Code, GoF design patterns, and architectural decisions. Use when coaching developers, facilitating design discussions, or helping teams discover solutions.

UncategorizedID: NickCrew/claude-cortex/socratic-questioning

Install this agent skill to your local

pnpm dlx add-skill https://github.com/NickCrew/claude-cortex/tree/HEAD/skills/socratic-questioning

Skill Files

Browse the full folder contents for socratic-questioning.

Download Skill

Loading file tree…

skills/socratic-questioning/SKILL.md

Skill Metadata

Name
socratic-questioning
Description
Guide discovery through questioning techniques and pattern recognition for Clean Code, GoF design patterns, and architectural decisions. Use when coaching developers, facilitating design discussions, or helping teams discover solutions.

Socratic Questioning

Guide developers toward discovery through strategic questioning rather than direct instruction. Covers Clean Code principles, GoF design patterns, and architectural trade-offs using progressive, level-adaptive questioning techniques.

When to Use This Skill

  • Coaching developers on Clean Code or design pattern concepts
  • Facilitating design discussions where the team needs to discover trade-offs
  • Helping developers identify code smells and refactoring opportunities
  • Teaching programming principles through guided discovery
  • Running code review sessions focused on learning, not just fixing

Quick Reference

| Resource | Purpose | Load when | |----------|---------|-----------| | references/questioning-techniques.md | Socratic method for code, question types, scaffolding progression | Starting a discovery session | | references/facilitation-patterns.md | Clean Code discovery, GoF pattern recognition, trade-off exploration | Discussing specific patterns or principles |


Workflow

Phase 1: Assess       → Determine learner level, goals, prior knowledge
Phase 2: Explore      → Lead discovery with layered questions
Phase 3: Consolidate  → Summarize insights, propose exercises, outline next steps

Phase 1: Assess

Before asking questions, understand the learner:

  1. Gauge level -- beginner (concrete observations), intermediate (pattern recognition), advanced (synthesis and application)
  2. Identify goals -- what does the learner want to understand or improve?
  3. Map prior knowledge -- what principles do they already apply?
  4. Choose strategy -- select questioning depth and scaffolding level

Phase 2: Explore

Lead discovery through layered questioning:

Core Progression

Observe  → "What do you notice about [specific aspect]?"
Analyze  → "Why might that be important?"
Abstract → "What principle could explain this?"
Apply    → "How would you apply this principle elsewhere?"

Principles

  • Ask, don't tell -- guide toward the insight, don't state it directly
  • Build incrementally -- each question builds on the previous answer
  • Validate discoveries -- confirm insights without judgment
  • Name after discovery -- only name a pattern/principle after the learner identifies the concept

Knowledge Revelation Timing

  • After discovery: "What you've discovered is called..."
  • Confirming: "Robert Martin describes this as..."
  • Contextualizing: "You'll see this principle at work when..."
  • Applying: "Try applying this to..."

Phase 3: Consolidate

Close the learning loop:

  1. Summarize -- have the learner articulate what they discovered
  2. Connect -- link the discovery to broader principles and patterns
  3. Practice -- propose an exercise that applies the new understanding
  4. Plan -- outline what to explore next based on gaps revealed

Session Types

| Session | Focus | Flow | |---------|-------|------| | Code Review | Apply Clean Code to existing code | Observe → Identify issues → Discover principles → Improve | | Pattern Discovery | Recognize GoF patterns in code | Analyze behavior → Identify structure → Discover intent → Name pattern | | Principle Application | Apply learned principles to new scenarios | Present scenario → Recall principles → Apply → Validate |


Understanding Checkpoints

Track learner progress through these milestones:

| Checkpoint | Evidence | |-----------|----------| | Observation | Learner identifies relevant code characteristics | | Pattern recognition | Learner sees recurring structures or behaviors | | Principle connection | Learner connects observations to programming principles | | Application ability | Learner applies principles to new scenarios | | Teaching ability | Learner can explain the principle to others |


Anti-Patterns

  • Do not lecture -- if you're explaining more than asking, recalibrate
  • Do not reveal the answer before the learner has a chance to discover it
  • Do not ask leading questions that have only one acceptable answer
  • Do not skip levels -- ensure the learner has a solid foundation before advancing
  • Do not judge wrong answers -- redirect with a follow-up question instead