Agent skill
skill-creator
Create new Claude Code skills with proper directory structure, SKILL.md file, and YAML frontmatter. Use this skill when you need to create a new reusable knowledge module for Claude Code.
Install this agent skill to your Project
npx add-skill https://github.com/d-oit/do-novelist-ai/tree/main/.claude/skills/skill-creator
SKILL.md
Skill Creator
Create new Claude Code skills following the official format and best practices.
Quick Reference
- Templates and Examples - Skill templates and complete creation examples
When to Use
- Creating a new reusable knowledge module
- Adding specialized guidance for specific tasks
- Building domain-specific expertise into Claude Code
- Need to ensure proper skill format and structure
Skill Structure
A Claude Code skill consists of:
.claude/skills/
└── skill-name/
└── SKILL.md
SKILL.md Format
---
name: skill-name
description: Clear description of what this skill does and when to use it (max 1024 chars)
---
# Skill Title
[Skill content in Markdown]
Naming Requirements
Skill Name Rules:
- Lowercase letters only
- Numbers allowed
- Hyphens for word separation (no underscores)
- No spaces
- Max 64 characters
- Descriptive and clear
Examples:
- ✅
episode-management - ✅
test-debugging - ✅
api-integration - ✗
Episode_Management(no uppercase, no underscores) - ✗
test debugging(no spaces)
Description Best Practices
The description is critical - Claude uses it to decide when to invoke the skill.
Good Description Structure:
[Action verb] [what it does] [when to use it]
Examples:
✅ Good:
description: Debug and fix failing tests in Rust projects. Use this skill when tests fail and you need to diagnose root causes, fix async/await issues, or handle race conditions.
✅ Good:
description: Implement new features systematically with proper testing and documentation. Use when adding new functionality to the codebase.
✗ Too vague:
description: Helps with testing
✗ Missing when-to-use:
description: Provides guidance on building APIs
Skill Creation Process
Step 1: Define Purpose
What problem does this skill solve?
- Specific task: [e.g., "Deploy to production"]
- Domain: [e.g., "deployment", "testing", "documentation"]
- User need: [e.g., "Ensure safe deployments"]
Step 2: Choose Name
Skill name: [lowercase-with-hyphens]
- Descriptive: Clearly indicates purpose
- Concise: Not too long
- Unique: Doesn't conflict with existing skills
Step 3: Write Description
description: [Action] [what it does]. Use this when [specific scenarios].
Key elements:
1. Clear action (verb)
2. What problem it solves
3. When to invoke it
4. Keywords Claude can match on
Step 4: Structure Content
Recommended Sections:
- Introduction: Brief overview of skill purpose
- When to Use: Specific scenarios for invocation
- Core Concepts: Key knowledge needed
- Process/Workflow: Step-by-step guidance
- Examples: Concrete usage examples
- Best Practices: Do's and don'ts
- Integration: How this works with other skills/agents
Content Guidelines:
- Clear, concise language
- Actionable instructions
- Concrete examples
- Code snippets where helpful
- Checklists for processes
- Visual diagrams (ASCII art) for complex flows
Step 5: Create Files
# Create directory
mkdir -p .claude/skills/skill-name
# Create SKILL.md with content
cat > .claude/skills/skill-name/SKILL.md << 'EOF'
---
name: skill-name
description: Your description here
---
# Skill Title
[Your skill content]
EOF
Step 6: Test and Validate
Validation Checklist:
- Directory name matches skill name
- SKILL.md file exists
- YAML frontmatter is valid
- Name follows naming rules (lowercase, hyphens)
- Description is clear and specific (< 1024 chars)
- Content is well-structured
- Examples are provided
- Markdown is properly formatted
Skill Templates
See templates-and-examples.md for complete templates including:
Available Templates
- Process Skill - Step-by-step workflows
- Knowledge Skill - Domain expertise and concepts
- Tool Skill - Tool usage and best practices
Each template includes:
- Full YAML frontmatter
- Recommended sections
- Example content
- Best practices structure
Integration with Agent Creator
When creating skills that work with agents:
- Reference agents in skill: Mention which agents use this skill
- Skill-agent coordination: Ensure skill complements agent capabilities
- Invocation clarity: Make clear when skill vs agent is appropriate
Project-Specific Considerations
For Rust Self-Learning Memory Project
Domain-Specific Skills:
- Episode management (start, log, complete)
- Pattern extraction and storage
- Memory retrieval optimization
- Turso/redb synchronization
- Async/Tokio patterns
Skill Naming Convention:
episode-[operation]for episode-related skillsstorage-[operation]for storage operationspattern-[operation]for pattern handlingmemory-[operation]for memory operations
Integration Requirements:
- Reference AGENTS.md standards
- Include examples using project structure
- Consider self-learning memory tracking
Skill Maintenance
Updating Skills
When updating existing skills:
- Preserve backward compatibility
- Update description if scope changes
- Add new sections without removing old ones
- Update examples to reflect current best practices
- Maintain clear version history in git
Deprecating Skills
If a skill becomes obsolete:
- Update description to indicate deprecation
- Point to replacement skill
- Keep file for backward compatibility
- Consider removing after transition period
Best Practices Summary
DO:
✓ Write clear, specific descriptions ✓ Include concrete examples ✓ Structure content logically ✓ Use consistent formatting ✓ Test skill by using it ✓ Update README.md to list new skill ✓ Follow naming conventions
DON'T:
✗ Use vague or generic descriptions ✗ Skip examples ✗ Make names too long or unclear ✗ Forget YAML frontmatter ✗ Use uppercase or underscores in names ✗ Exceed 1024 chars in description
Validation Command
After creating a skill, validate it:
# Check structure
test -f .claude/skills/skill-name/SKILL.md && echo "✓ Structure correct"
# Check YAML frontmatter
head -n 5 .claude/skills/skill-name/SKILL.md | grep "^name:" && echo "✓ YAML valid"
# Check name format
[[ $(grep "^name:" .claude/skills/skill-name/SKILL.md | cut -d' ' -f2) =~ ^[a-z0-9-]+$ ]] && echo "✓ Name format correct"
Quick Creation Script
#!/bin/bash
# create-skill.sh
SKILL_NAME=$1
DESCRIPTION=$2
if [ -z "$SKILL_NAME" ] || [ -z "$DESCRIPTION" ]; then
echo "Usage: ./create-skill.sh skill-name \"Skill description\""
exit 1
fi
# Validate name format
if ! [[ "$SKILL_NAME" =~ ^[a-z0-9-]+$ ]]; then
echo "Error: Skill name must be lowercase with hyphens only"
exit 1
fi
# Create directory
mkdir -p ".claude/skills/$SKILL_NAME"
# Create SKILL.md
cat > ".claude/skills/$SKILL_NAME/SKILL.md" <<EOF
---
name: $SKILL_NAME
description: $DESCRIPTION
---
# ${SKILL_NAME^}
[Skill content goes here]
## When to Use
- [Scenario 1]
- [Scenario 2]
## Process
### Step 1: [Action]
[Instructions]
### Step 2: [Action]
[Instructions]
## Examples
### Example 1: [Name]
\`\`\`
[Example code or workflow]
\`\`\`
## Best Practices
✓ [Do this]
✗ [Don't do this]
EOF
echo "✓ Created skill: $SKILL_NAME"
echo "✓ Edit: .claude/skills/$SKILL_NAME/SKILL.md"
Summary
Creating effective skills:
- Purpose: Solve specific, well-defined problems
- Naming: Clear, lowercase, hyphenated names
- Description: Specific, actionable, includes when-to-use
- Structure: Well-organized with clear sections
- Examples: Concrete, realistic usage examples
- Testing: Validate structure and use the skill
Skills are the foundation of Claude Code's knowledge. Well-designed skills make Claude more effective at autonomous task execution.
For complete templates and detailed examples, see templates-and-examples.md.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
novel-development
Work on novel-specific features including plot engines, character development, world-building, timeline management, and GOAP-based story generation. Use when implementing narrative systems, character arcs, or story planning tools.
tech-stack-specialist
Manage framework usage, dependencies, build configuration, and environment setup. Use when adding new dependencies, updating packages, configuring build tools, or setting up development environment.
performance-engineer
Optimize application performance including build times, runtime speed, bundle size and resource usage. Use when addressing performance issues, implementing caching strategies, or optimizing rendering.
e2e-test-optimizer
Optimize Playwright E2E tests by removing anti-patterns, implementing smart waits, enabling test sharding, and improving reliability.
domain-expert
Apply domain-driven design principles for business logic, entities, events and aggregate boundaries. Use when modeling domain concepts, implementing business rules, or defining clear separation between domain and infrastructure layers.
qa-engineer
Define comprehensive testing strategies, write tests with proper naming conventions, organize tests by type, and implement mocking strategies. Use when creating tests, refactoring test suites, or improving test coverage.
Didn't find tool you were looking for?