女娲.skill
Turns the thinking styles of famous people into AI agent skills you can call on.alchaincyf/nuwa-skill
- Stars
- 37.2k
- Forks
- 4.8k
- Licence
- MIT
- Language
- Python
- Last update
- 12 days ago
- Type
- AI model / agent
What it is
Nuwa (女娲.skill) is an Agent Skills package that, given a person's name, researches their public work and writes a SKILL.md capturing how they tend to think, decide, and talk. Once installed in a supported AI agent, you can ask the agent to answer questions or analyze problems from that person's perspective. The README says it distills mental models, decision heuristics, and speaking style, and notes limits such as relying only on public information.
- Ask a distilled perspective (e.g. Munger) to analyze an investment decision
- Ask how Feynman would explain a technical topic such as quantum computing
- Generate a new perspective skill for a person by telling the agent to distill them by name
- Compare how different distilled figures respond to the same business question
Great for
- People who want an AI assistant to reason through a decision using a known thinker's frameworks
- Students exploring how figures like Feynman or Munger approach problems
- Developers already using Claude Code, Codex, Cursor, or another Agent Skills-compatible runtime
- People who want to build their own perspective skill for a person not in the list
Not ideal for
- Anyone wanting factual, verified statements from the real person: the README says public expression is not the same as real views, and skills are snapshots as of research time
- Users whose agent runtime does not support Agent Skills (you would have to paste SKILL.md manually, which the README calls a reference-only option)
- People looking for a conventional software library or app; this is a set of prompt-based skill files
How to get it
- 1
Install with a package manager
You'll need: An Agent Skills-compatible AI agent runtime (e.g. Claude Code, Codex, Cursor) · Node.js (needed for the npx command shown in the README)
Install via the universal skills CLI
npx skills add alchaincyf/nuwa-skill - 2
Run it
- Install the skill with the npx command or by asking your agent to install it from the repo URL
- Open your agent and tell it to distill a person, e.g. 蒸馏一个保罗·格雷厄姆
- Wait for the agent to finish research and build the new skill
- Invoke the skill by name, e.g. 用芒格的视角帮我分析这个投资决策
- 3
Get the most out of it
- Use the examples/ directory to see full research data and sample conversations for included figures
- Read references/extraction-framework.md to understand how mental models are selected and validated
- Check the per-skill FIDELITY.md files to compare quality before relying on a persona
- Use the companion darwin-skill project, which the README says applies an evaluation loop to refine skills
Downloads
Strengths
- Built on the open Agent Skills standard and says it works across many runtimes
- Includes a stated honesty section listing what each skill cannot do
- Ships with example conversations and a published fidelity scorecard method in the repo
Trade-offs
- Output quality depends on the underlying AI agent and the public material available for each person
- The fidelity scores are self-reported by the project and not independently verified here
- Hype-prone framing and a large set of named people may make it hard to judge individual skills
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.