huashu-nuwa

Generates persona perspective skills by researching public figures and extracting their mental models.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/composable-tu/skills-test --skill huashu-nuwa-composable-tu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: huashu-nuwa
Source: https://github.com/composable-tu/skills-test/tree/main/.trae/skills/huashu-nuwa
Command: npx skills add https://github.com/composable-tu/skills-test --skill huashu-nuwa-composable-tu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires yt-dlp, and includes scripts (resource) and references (resource) components.

What problem does it solve? It automates the creation of AI persona skills that replicate how a public figure thinks, turning a single name into a researched, validated, runnable perspective skill instead of shallow role-play prompts. ## Core Features & Use Cases - Six-stream parallel research: Spawns agents covering writings, interviews, social media, critic views, decision records, and life timeline, with source blacklists and credibility tagging. - Triple-verified extraction: Distills 3-7 mental models, 5-10 decision heuristics, expression DNA, values, and honest limits using cross-domain, generative, and exclusivity tests. - Quality validation pipeline: Runs sanity, edge-case, and voice tests plus automated quality_check.py scoring before delivery. - Use Case: Say "distill Munger" and receive a complete SKILL.md that answers investment questions using Munger's actual cognitive frameworks, with explicit honesty boundaries. ## Quick Start Ask the assistant to distill a named public figure into a perspective skill, for example by saying "distill Paul Graham into a thinking advisor skill".

Frequently Asked Questions about huashu-nuwa

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create an AI persona skill of a public figure?

Provide a name and the skill runs six parallel research agents covering writings, interviews, social media, critics, decisions, and timeline. It then extracts verified mental models, decision heuristics, and expression DNA into a ready-to-use SKILL.md file.

How to distill someone's thinking style into a Claude Code skill?

Install the skill and say "distill X" with any public figure's name. The workflow researches primary sources, validates mental models through triple verification, builds the SKILL.md from a template, and runs quality tests before delivery.

Can I use my own materials instead of web research?

Yes, local corpus mode accepts user-provided PDFs, transcripts, subtitles, and blog exports as highest-priority sources. Web search then only fills dimensions your local materials do not cover, or is skipped entirely for private individuals.

What are the limitations of AI persona skills?

They cannot replicate intuition or creativity, only capture a snapshot up to the research date, and rely on public statements which may differ from private beliefs. Every generated skill explicitly lists these honest boundaries.

Why does persona skill generation produce generic results?

Generic output usually means too few primary sources or models that fail the exclusivity test. The workflow counters this with source blacklists, a 50% primary-source requirement, and a quality_check.py script that scores six pass criteria.