myclaude-creator-engine

---

Thousands of Claude Code users build custom skills, agents, and tools every day. But there is no standard for quality — no way to know if what you built is robust, no guided process to follow, and no easy way to share your work with others.

The Studio Engine is the missing creation pipeline. It walks you from idea to published product: researching your domain, generating structure, guiding you through content, scoring quality against 20 structural patterns, and publishing to a marketplace where anyone can install your work with one command.

You don't need to write code. You need expertise worth packaging.

---

Install in 30 Seconds

myclaude studio

Don't have the CLI yet? npm i -g @myclaude-cli/cli — then run the command above.

Or clone directly: git clone https://github.com/myclaude-sh/myclaude-creator-engine && cd myclaude-creator-engine && claude

That's it. Open the folder in Claude Code and run /onboard — the engine adapts to you.

---

See It in Action

A Kubernetes security consultant used the engine to turn 15 years of expertise into an installable advisor — in one sitting:

/scout kubernetes-security

The engine tested what Claude already knows about K8s security, found 11 gaps it can't cover alone, scanned the marketplace for existing tools, and recommended building a cognitive mind.

/create minds

Generated a product structure with guided annotations — 7 sections, each with context about what belongs there and why.

/fill

The engine asked targeted questions: "What's the first thing you check in a cluster audit?" "Walk me through an attack path from exposed etcd." The consultant answered. The engine wrote the structure.

/validate

Scored 100% — Elite quality. 20/20 structural patterns passing. Attack-path reasoning, CIS benchmarks, hardening checklists — all verified.

/publish

Live on myclaude.sh. Anyone in the world can now run: myclaude install k8s-security-advisor and Claude Code gains deep K8s security expertise it doesn't have natively.

The consultant provided the expertise. The engine handled everything else.

✦ That's the thesis: Claude is powerful out of the box. But your Claude — augmented with real expertise — is something else entirely.

---

What You Can Build

You are a...	You build...	Start with
Developer	Skills, agents, hooks, multi-agent squads	/create skill
Consultant / Coach	Methodology frameworks anyone can install and use	/create workflow
Domain expert	Installable knowledge advisors with deep reasoning	/create minds
Team lead	Shared multi-agent teams with quality standards	/create squad
Writer / Researcher	Knowledge tools that augment Claude's capabilities	/scout your-domain
Anyone with expertise	Whatever Claude Code can't do alone	/scout then /create

Every product is self-contained — no engine dependency after install. Portable, yours, forever.

See more scenarios: Use Cases — lawyers, marketers, consultants, solo businesses, researchers.

Not sure what to build? Run /scout marketing-strategy (or any domain). The engine tests what Claude already knows, finds the gaps, and recommends exactly what to build. You start with intelligence, not guesswork.

---

How It Works

  research       create       refine       verify       ship
  /scout    →   /create  →   /fill   →  /validate →  /publish
                   │                        │
                   │    AI-guided filling    │  20 structural
                   │    with your expertise  │  quality patterns
                   └────────────────────────┘

v3.0.0 expanded validation from 7 stages to 10 — adding intent coherence (does the product match what the creator set out to build?), voice coherence (is the tone consistent?), and value intelligence (market positioning and pricing signals).

The engine asks the right questions. You provide the expertise. Each step feeds the next — scout research flows into /create, which generates structure for /fill, which produces content scored by /validate. Nothing is disconnected.

Phase	Command	What It Does
Profile	/onboard	Set up your creator profile — the engine adapts to you
Research	/scout [domain]	Test Claude's knowledge, find gaps, scan competition, recommend what to build
Knowledge	/map [topic]	Extract and structure your domain expertise into a reusable map
Create	/create [type]	Generate a product with structural DNA and guided annotations
Fill	/fill	AI-guided content filling — the engine asks, you answer, it writes
Quality	/validate	Score against 20 structural patterns (Verified / Premium / Elite)
Test	/test	Behavioral validation in an isolated sandbox — 3 scenarios
Package	/package	Strip annotations, generate manifests, prepare for distribution
Publish	/publish	Ship to myclaude.sh with one command
Import	/import	Bring existing skills into the pipeline for validation and publishing
Status	/status	Dashboard — scores, next steps, portfolio overview
Help	/help	Command reference with personalized recommendations
Think	/think	Brainstorm and compare approaches before committing
Explore	/explore	Search the marketplace for existing tools and inspiration
Security	/aegis	Security audit for any codebase — works independently

---

What Makes This Different

Other tools in the Claude Code ecosystem give you pre-built configurations — collections of skills, agents, and rules to install. They are warehouses of ready-made parts.

The Studio Engine is a factory. It helps you create your own.

	Pre-built config tools	Studio Engine
What you get	Someone else's skills	Your own products
Quality	Trust the author	Scored against 20 patterns
Process	Copy and install	Research → create → validate → publish
Audience	Developers only	Anyone with expertise
Distribution	GitHub repo	Marketplace with one-command install
After install	Depends on the repo	Self-contained, no dependency

No other tool in the ecosystem validates quality before publishing. Your products are verified, not just functional.

---

Quality You Can Measure

Every product is scored: (DNA x 0.50) + (Structural x 0.30) + (Integrity x 0.20)

Tier	Score	What It Means
Verified	>= 75%	Functional, documented, core patterns present
Premium	>= 85%	Professional craft — advanced structural patterns
Elite	>= 92%	State-of-the-art — deep structural quality

20 structural patterns across 10 validation stages check for: activation protocols, error handling, anti-pattern guards, progressive disclosure, quality gates, graceful degradation, cache-friendly design, and more.

Run /validate at any time. The engine shows exactly what passes, what needs work, and how to fix it — with specific instructions, not vague suggestions.

---

The Ecosystem

         YOU                          ANYONE
          │                             │
     ┌────┴────┐                   ┌────┴────┐
     │  BUILD  │                   │ INSTALL  │
     │ Engine  │  ──── publish ──→ │   CLI    │
     │ /create │                   │ myclaude │
     │ /fill   │                   │ install  │
     │/validate│                   │          │
     └─────────┘                   └──────────┘
          │                             │
          └──── myclaude.sh ────────────┘
                marketplace

Component	What It Is	Link
Engine (this repo)	Create and validate products	MIT, open source
CLI (@myclaude-cli/cli)	Search, install, publish from the terminal	myclaude.sh
Marketplace	Browse and install with one command	myclaude.sh

Products created by the engine use the Agent Skills format — YAML frontmatter with name, description, and constraints, followed by markdown instructions. This is the same SKILL.md format used natively by Claude Code and recognized by other AI coding tools that support structured skill definitions.

---

Each type has specific structural DNA — patterns that ensure it works correctly, handles edge cases, and activates reliably.

Type	What It Is	Best For
skill	Single-purpose tool with focused capability	Code review, docs generation, audits, analysis
agent	Autonomous multi-step task runner	Complex automation with decision-making
squad	Multi-agent team with intelligent routing	Multi-domain tasks where agents coordinate
workflow	Orchestrated process automation	Standardized team processes and methodologies
minds	Domain expert advisor (advisory or cognitive depth)	Packaging deep knowledge as installable intelligence