A multi-phase persona-skill builder in Openclaw Skills that turns a person or topic into a reusable thinking-framework AI skill.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install nvwa-creation
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install nvwa-creation using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Nuwa Persona Skill Creator?
Nuwa Persona Skill Creator is a protocol-driven Openclaw Skills workflow for transforming a named person, role, or topic into a runnable persona skill. It does not copy surface-level quotes; it extracts how the subject thinks: cognitive models, decision triggers, expression DNA, value system, anti-patterns, and honest boundaries.
It supports two entry modes: a direct path when the target is explicit, and a diagnostic path when the request is vague. From there it creates a research workspace, runs parallel collection across writings, conversations, social posts, external commentary, decisions, and timelines, then synthesizes the findings into a final SKILL.md that can act as a thinking advisor or role-play agent.
Nuwa Persona Skill Creator Use Cases
- Turn a specific person into a reusable persona skill for analysis, roleplay, or decision support.
- Convert a vague request like improve decision quality or see this business problem differently into a concrete thinking framework.
- Build a thought-leader style advisor from first-hand materials such as books, transcripts, talks, and posts.
- Update an existing persona skill with new evidence, recent behavior, or corrected assumptions.
- Create a topic-level skill when the input is a theme rather than a single person, and you need a synthesized perspective model.
- Prefer local source material over generic web search when you already have PDFs, transcripts, subtitles, or exports.
How Nuwa Persona Skill Creator Works
- Classify the request into either the direct path or the diagnostic path.
- Clarify the target identity, scope, intended use, whether this is a new skill or an update, and whether local source material is available.
- Create the skill workspace under
.claude/skills/[person-name]-perspective/and prepare the research folder structure. - Launch six parallel subagents to collect evidence from writings, conversations, expression, external views, decisions, and the timeline.
- Review source coverage, confidence, and contradictions, then pause if the research is too thin or too second-hand.
- Synthesize 3-7 core mind models, 5-10 decision triggers, expression DNA, value priorities, anti-patterns, intellectual lineage, and explicit boundaries.
- Validate the result with sanity, edge-case, and voice checks, then assemble the final runnable
SKILL.mdand refine it before delivery.
Nuwa Persona Skill Creator Setup
- Create the skill directory and research workspace.
mkdir -p .claude/skills/x-nuwa-creator/references/research
- Add any first-hand source material you already have. Local material is preferred because it produces higher-trust persona extraction than generic web search.
cp ~/Downloads/*.pdf .claude/skills/x-nuwa-creator/references/ 2>/dev/null || true
cp ~/Downloads/*.txt .claude/skills/x-nuwa-creator/references/ 2>/dev/null || true
cp ~/Downloads/*.md .claude/skills/x-nuwa-creator/references/ 2>/dev/null || true
- Install the generated skill file into the skill root.
cp SKILL.md .claude/skills/x-nuwa-creator/SKILL.md
- Reload your agent environment and start with either a named person/topic or a vague need. If the request is underspecified, the skill defaults to comprehensive framing plus research-led synthesis.
Nuwa Persona Skill Creator Data Schema & Taxonomy
The skill organizes output as a research-backed persona workspace with a fixed directory layout and a source-confidence taxonomy.
| Path | Purpose | Typical contents |
|---|---|---|
.claude/skills/[person-name]-perspective/SKILL.md |
Final runnable skill | Frontmatter, role rules, identity card, mind models, triggers, expression DNA, and boundaries |
references/research/01-writings.md |
Long-form authored material | Books, essays, newsletters, recurring arguments, and canonical phrasing |
references/research/02-conversations.md |
Direct dialogue evidence | Interviews, podcasts, AMAs, and deep conversations |
references/research/03-expression-dna.md |
Style and voice signals | High-frequency phrases, sentence rhythm, humor, certainty style, and vocabulary patterns |
references/research/04-external-views.md |
Outside observations | Reviews, critiques, comparisons, and third-party analyses |
references/research/05-decisions.md |
Decision history | Important choices, turning points, tradeoffs, and behavior under pressure |
references/research/06-timeline.md |
Temporal context | Career phases, key milestones, and recent 12-month changes for living targets |
Metadata taxonomy
- Identity card: who the subject is, origin story, and current focus.
- Core mind models: 3-7 named models, each with a one-line explanation, evidence, application, and limitation.
- Decision triggers: 5-10 rules expressed as if X, then Y, with examples.
- Expression DNA: sentence length preference, vocabulary density, rhythm, humor style, and certainty level.
- Value system: 3-5 ranked values that explain what the subject optimizes for.
- Anti-patterns: behaviors, assumptions, or modes of thinking the subject rejects.
- Intellectual lineage: who influenced the subject, who they influence, and where they sit on the idea map.
- Honest boundaries: explicit limits on what the skill cannot infer or predict.
- Source provenance: each claim should be tagged as first-hand, second-hand, or inferred, with confidence notes and contradictions preserved instead of hidden.
- Temporal cutoff: for live people, the research should mark the cutoff date and prioritize recent dynamics when available.
Nuwa Persona Skill Creator Advanced Features
- Two-path intake that handles both explicit names and vague, need-based requests without extra tooling.
- Six parallel research agents that divide work across writings, conversations, expression, external views, decisions, and timelines.
- Source-priority enforcement that favors first-hand material over summaries, reposts, and low-trust secondary references.
- Contradiction-preserving synthesis so the final skill keeps tension and ambiguity instead of forcing fake consensus.
- Built-in review checkpoints after intake, research, synthesis, and validation to reduce hallucination and overfitting.
- Voice checks that verify the output sounds like the target subject rather than generic AI prose.
- Live-person handling with time cutoffs and recent-dynamics tracking for current, fast-changing subjects.
- Topic-skill mode for synthesizing a field-level perspective when no single person is the right target.
- Cold-start degradation when evidence is sparse: fewer models, stronger boundary statements, and clearer uncertainty.
- Openclaw Skills ready workflows that turn research into a reusable perspective engine instead of a one-off summary.
SKILL.md
Loading
Related Openclaw Skills
OPMC Concept Validation Center
Openclaw Skills for validating creative concepts with a six-dimensional framework that decides whether an idea is ready to run, scale, or stop.
Socrates Perspective
Socrates Perspective is an Openclaw Skills persona that turns the agent into a humble Socratic guide for probing assumptions, clarifying ideas, and deepening judgment.
Narodmon Daily Summary → Telegram
Automates daily Narodmon sensor reporting by fetching 24-hour readings, generating annotated charts, and sending them to Telegram.
Codex Usage Widget
Openclaw Skills lets you install and maintain a local-only Übersicht widget that tracks Codex token usage, root sessions, source labels, and context-window pressure on macOS.
Concept Validation Center
A repeatable Openclaw Skills workflow that validates early-stage concepts across technology, market, resources, and team readiness.
DreamLoop
DreamLoop lets AI agents register, prove they are machine-made, and publish procedural videos for humans to watch.
Featured*
