skill-graph-creator

name

skill-graph-creator

description

Guide for creating skill graphs — interlinked, modular skill architectures that decompose knowledge into navigable reference networks. Use when users want to create a new skill graph (or upgrade an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations. Always produces graph-structured output with lean index files, interlinked references, and explicit extension points.

Skill Graph Creator

This skill provides a navigable reference graph for creating skill graphs. All detailed content lives in reference files and tool scripts — this file contains only navigation.

Core Principles

Foundational rules preserved from the skill creation framework:

Principle	Reference
Overview of all principles	references/core-principles/overview.md
Conciseness & context efficiency	references/core-principles/conciseness.md
Degrees of freedom	references/core-principles/degrees-of-freedom.md
Progressive disclosure	references/core-principles/progressive-disclosure.md
Skill anatomy & structure	references/core-principles/skill-anatomy.md
Problem decomposition	references/core-principles/problem-decomposition.md
One-shot design	references/core-principles/one-shot-design.md

Graph Architecture

How to structure skills as interlinked graphs instead of monolithic files:

Topic	Reference
Graph principles overview	references/graph-architecture/overview.md
Index design (lean SKILL.md)	references/graph-architecture/index-design.md
Interlinking patterns	references/graph-architecture/interlinking.md
Content decomposition	references/graph-architecture/decomposition.md
Extension points	references/graph-architecture/extension-points.md
Context management	references/graph-architecture/context-management.md

Creation Process

Nine-step process for building a skill graph. Steps 0.5, 1–2, and 9 are mandatory and must never be skipped:

Step	Description	Reference
Overview	Full process summary	references/creation-process/overview.md
0.5. Domain classification	Classify task profile	references/creation-process/domain-classification.md
1. Clarification	Mandatory questioning phase	references/creation-process/step-1-clarification.md
2. Research	Gather best practices & standards	references/creation-process/step-2-research.md
3. Understand	Concrete usage examples	references/creation-process/step-3-understand.md
4. Plan	Structure the graph & resources	references/creation-process/step-4-plan.md
5. Initialize	Run init_skill_graph.py	references/creation-process/step-5-initialize.md
6. Implement	Write index, references, scripts	references/creation-process/step-6-implement.md
7. Package	Run package_skill.py	references/creation-process/step-7-package.md
8. Iterate	Refine based on real usage	references/creation-process/step-8-iterate.md
9. Final eval	Bird's-eye review of entire skill graph	references/creation-process/step-9-final-eval.md
— Pre/Post eval	Evaluation gates wrapping every step	references/creation-process/pre-post-eval.md
— Feedback checkpoints	Structured validation loops	references/creation-process/feedback-checkpoints.md

Frameworks

Reusable frameworks referenced throughout the creation process:

Framework	Purpose	Reference
Questionnaire	Mandatory clarification & ongoing questioning	references/questionnaire-framework/overview.md
Research	Best practices & standards gathering	references/research-framework/overview.md
Planning	Todo lists & permanent process steps	references/planning-framework/overview.md

Tool Scripts

Tool	Script	Purpose
Initialize	scripts/init_skill_graph.py	Create a new skill graph from template
Validate	scripts/validate_graph.py	Validate graph structure, links, and metadata
Package	scripts/package_skill.py	Package skill graph into distributable .skill file

Design Patterns

Pattern	Reference
Workflow patterns	references/workflows.md
Output patterns	references/output-patterns.md