What problem does it solve? It turns the vague wish to think like someone such as Munger, Feynman, or Musk into a concrete, validated Claude Code skill, automating the research, extraction, and quality-checking work that manual persona building requires. ## Core Features & Use Cases - Six-Agent Parallel Research: Spawns agents covering writings, interviews, social media, critic views, decision records, and life timeline, each archiving sourced findings to disk. - Triple-Verified Extraction: A claim becomes a mental model only if it recurs across 2+ domains, predicts new positions, and is non-generic; results are assembled into 3-7 mental models, decision heuristics, expression DNA, anti-patterns, and honest limits. - Quality Gates and Refinement: Includes research review checkpoints, sanity/edge/voice tests via subagents, automated quality_check.py scoring, and a dual-agent refinement pass. - Use Case: Say "distill Paul Graham" and receive a self-contained perspective skill directory with SKILL.md, research files, and helper scripts, ready to invoke as a thinking advisor. ## Quick Start Ask the assistant to distill a named person or topic into a perspective skill, for example by saying you want a Munger-style thinking advisor built from his public writings and talks.
