Agent skill
skill-creator
Create and manage Claude Code skills. Use when creating a new skill, updating an existing skill, or debugging skill activation issues. Covers skill structure, skill-rules.json configuration, trigger patterns, and best practices.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/skill-creator-bliz243-dotfiles
SKILL.md
Skill Creator
Guide for creating effective Claude Code skills.
What Skills Are
Skills are modular packages that extend Claude's capabilities with specialized knowledge, workflows, and tools. They transform Claude from a general-purpose agent into a domain expert.
Default assumption: Claude is already smart. Only add context Claude doesn't have. Challenge each piece: "Does this justify its token cost?"
Skill Structure
skill-name/
├── SKILL.md # Required - main instructions
├── references/ # Optional - detailed docs loaded on-demand
├── scripts/ # Optional - executable code
└── assets/ # Optional - templates, images, files for output
SKILL.md (Required)
---
name: my-skill
description: What it does AND when to use it. Include trigger keywords.
---
# My Skill
## Quick Start
[Essential workflow - what to do first]
## Patterns
[Show correct patterns with code examples]
## Anti-Patterns
[Show what to avoid with code examples]
## Reference
- See [detailed-ref.md](references/detailed-ref.md) for more
Key rules:
- Keep under 500 lines (use references/ for details)
- Description is the trigger - include all "when to use" info there
- Show examples over explanations
- Imperative form: "Use X" not "You should use X"
References (Optional)
For detailed docs loaded only when needed:
references/
├── patterns.md # Detailed code patterns
├── api-docs.md # API reference
└── examples.md # More examples
Keep references one level deep. For large files (>100 lines), add a table of contents.
Skill Rules Configuration
Add to ~/.claude/config/skill-rules.json to control when skills activate:
{
"skills": [
{
"name": "my-skill",
"type": "suggest",
"triggers": ["keyword1", "keyword2"],
"intentPatterns": ["(create|add).*?component"],
"description": "Shown when skill is suggested"
}
]
}
Fields
| Field | Required | Description |
|---|---|---|
name |
Yes | Skill name (matches folder name) |
type |
Yes | suggest (advisory) or block (required) |
triggers |
No | Keywords that activate skill (case-insensitive) |
intentPatterns |
No | Regex patterns for implicit intent |
description |
No | Message shown when suggested |
Trigger Types
Keywords - exact substring match (case-insensitive):
"triggers": ["aggregate", "repository", "svelte"]
- "create aggregate" ✅
- "Create Aggregate" ✅ (case-insensitive)
- "aggregation" ✅ (substring)
Intent Patterns - regex for action + domain:
"intentPatterns": [
"(create|add|implement).*?(service|component)",
"(pause|cancel|resume).*?subscription"
]
Tips:
- Use non-greedy:
.*?not.* - Common verbs: create, add, implement, modify, fix, build
- Test at regex101.com
Examples
Simple Skill (No Rules)
Just create the folder with SKILL.md:
~/.claude/skills/code-review/
└── SKILL.md
Invoke with /code-review or skill triggers from description.
Skill with Auto-Suggestion
{
"skills": [
{
"name": "frontend-design",
"type": "suggest",
"triggers": ["ui", "component", "styling", "css", "tailwind"],
"intentPatterns": ["(build|create|design).*?(page|component|ui)"],
"description": "Use frontend-design for distinctive UI"
}
]
}
Domain-Specific Skill
ddd-patterns/
├── SKILL.md # Core workflow
└── references/
├── aggregates.md # Aggregate patterns
├── repositories.md # Repository patterns
└── services.md # Service patterns
SKILL.md references these only when relevant:
## Aggregates
See [references/aggregates.md](references/aggregates.md) for patterns.
## Repositories
See [references/repositories.md](references/repositories.md) for patterns.
Best Practices
Do:
- Keep SKILL.md under 500 lines
- Show concrete code examples
- Use references/ for detailed docs
- Make triggers specific to avoid false positives
- Test trigger patterns before deploying
Don't:
- Over-explain concepts (show patterns instead)
- Use broad keywords ("system", "work", "file")
- Include README, CHANGELOG, or meta-docs
- Duplicate info between SKILL.md and references
Debugging
Skill Not Triggering
- Check skill folder exists in
~/.claude/skills/ - Verify SKILL.md has valid frontmatter (name + description)
- Check skill-rules.json is valid JSON:
jq . ~/.claude/config/skill-rules.json - Test keywords are in your prompt (case-insensitive substring)
- Test intent patterns at regex101.com
Creating a New Skill
- Create folder:
mkdir ~/.claude/skills/my-skill - Create SKILL.md with frontmatter and instructions
- Optionally add to skill-rules.json for auto-suggestion
- Test with
/my-skillor trigger keywords - Iterate based on real usage
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?