Problem complexity classification and learning progression framework with quantitative wickedness assessment methodology. Features 5 problem tiers (Simple→Super-Wicked), Base-N architecture (Base6→BASE120), and empirical validation from HUMMBL mental models research. Includes automated quality control and community templates.

Verification-graded analysis for high-stakes, complex, and wicked problems — a Claude Code and Claude Cowork plugin: source-tiered web verification, a disconfirmation pass, grade-locked decision briefs, and audience rendering, architected around the ICD-203 analytic tradecraft standards.

My Bachelor's thesis application for visualizing wicked problems with real-time multi-user support.

Some foundational documents

🤖 Official code for "The LLM Team Composition Paradox." This research challenges the "complexity = better" assumption by proving the optimal LLM agent team (single, homogeneous, or diverse) is contingent on the problem type, implementing a Hierarchical OODA-Belbin framework.

A multi-agent system for dynamic mapping of wicked problems in AI governance

correlating resonance is a methodology that clusters concepts used in a corpora of documents to identify frames of reference in conflicted policy spaces