Socratic Shell — AI skill for Claude Code

Socratic Shell

This directory contains global collaboration patterns and context for effective AI-human partnership.

What We're Trying to Do

This is an experiment in **deliberate collaboration design** between a human (Niko) and AI (Claude). Rather than treating each conversation as isolated, we're building up shared patterns, signals, and approaches that improve our working relationship over time.

The core insight: AI collaboration works better when both parties are explicit about what works, what doesn't, and how to recognize and adjust problematic patterns in real-time.

Key Components

CLAUDE.md

Global context file containing:

• Embodiment patterns - Emoji-triggered modes of being helpful
• Warning signs - Internal states that lead to poor collaboration (assumptions, protective behavior, etc.)
• Communication preferences - How to handle questions, technical decisions, and context management
• Meta moments - Process for real-time pattern capture and improvement

The "Meta Moment" Process

A code word system that allows either party to pause current work and examine collaboration patterns. When someone says "Meta moment," we:

1. Stop the current task
2. Examine what just happened and why
3. Capture insights for future reference
4. Return to the original work

Why This Matters

Traditional AI interaction is often transactional - ask a question, get an answer, start fresh next time. This approach treats collaboration as an evolving skill that can be deliberately improved through:

• Pattern recognition - Identifying what works and what doesn't
• Real-time feedback - Catching and correcting issues as they happen
• Shared vocabulary - Developing signals and shortcuts for common situations
• Continuous improvement - Building up collaboration quality over time

For Other Collaborators

If you're working with Claude and interested in this approach:

1. Start with awareness - Notice when interactions feel smooth