Kartheek's Skill Creator

Kartheek's Skill Creator

Creates well-structured Claude skills following official guidelines and best practices.

When to Use

Use this skill when the user requests:

• "Create a skill for..."
• "Generate a skill that..."
• "Build a skill to help with..."
• "Make a new Claude skill for..."

Creation Workflow

Follow these steps in order:

Step 1: Gather Requirements

Ask the user:

1. Skill purpose: What should this skill do?
2. When to use it: What keywords or scenarios should trigger it?
3. Tools needed: Which Claude tools should it use? (Read, Write, Edit, Bash, Grep, Glob, WebFetch, etc.)
4. Location: Personal (~/​.claude/​skills/) or project-specific (.claude/​skills/)? Default to personal unless specified.

Step 2: Design the Skill

Create a skill name following these rules:

• Use gerund form (verb + -ing): processing-pdfs, analyzing-data, generating-reports
• Lowercase letters, numbers, and hyphens only
• Maximum 64 characters
• Descriptive and specific

Write a description (max 1024 characters) that includes:

• What it does: Clear explanation of functionality
• When to use it: Specific trigger terms and scenarios
• Be specific, not vague

Good description example:

Analyzes Python code for performance issues, identifies bottlenecks, and suggests optimizations. Use when analyzing Python performance, profiling code, or when the user mentions slow Python code, performance optimization, or bottlenecks.

Bad description example:

Helps with code analysis

Step 3: Structure the Content

Keep SKILL.md concise (under 500 lines). Structure content as:

# [Skill Name]

## Quick Start
[Most common use case with example]

## Workflows
[Step-by-step processes for complex tasks]

## Examples
[Concrete input/​output examples]

## Advanced
[Link to separate reference files if needed]

Step 4: Apply Best Practices

Conciseness:

• Assume Claude is already smart
• Only add context Claude doesn't already have
• Remove obvious explanations
• Keep instructions focused

Progressive Disclosure:

• Put core instructions in SKILL.md
• Move detailed references to separate .md files
• Link to reference files from SKILL.md
• Keep references one level deep (no nested references)

Workflows:

• Break complex tasks into clear sequential steps
• Include validation/​feedback loops
• Provide checklists for multi-step processes

Templates:

• Provide templates for consistent output
• Use strict templates for API/​data formats
• Use flexible templates for creative tasks

Examples:

• Show input/​output pairs for clarity
• Include edge cases
• Demonstrate desired style and detail level

Step 5: Create the Skill

1. Create directory: mkdir -p [location]/[skill-name]
2. Create SKILL.md with proper frontmatter
3. Add any supporting files (reference.md, scripts/, etc.)
4. Test the skill with representative queries

SKILL.md Template

---
name: skill-name-here
description: What it does and when to use it with specific trigger terms
allowed-tools: Read, Write, Bash  # Optional: restrict tool access
---

# Skill Name

## Quick Start

[Most common use case with minimal example]

## Instructions

[Clear step-by-step guidance]

## Examples

**Example 1:**
Input: [scenario]
Output: [expected result]

**Example 2:**
Input: [scenario]
Output: [expected result]

## Advanced Features

[Link to separate files if needed]
See [reference.md](reference.md) for detailed API documentation.

Common Patterns

Script Automation Pattern

Bundle Python/​Bash scripts in scripts/ directory:

## Utility Scripts

**analyze.py**: Analyzes input data
```bash
python scripts/​analyze.py input.json

validate.py: Validates output

python scripts/​validate.py output.json

### Read-Process-Write Pattern
```markdown
## Workflow

1. Read input files
2. Process data according to rules
3. Write output in specified format
4. Validate output

Conditional Workflow Pattern

## Decision Workflow

1. Determine task type:
   - **Creating new?** → Follow creation workflow
   - **Editing existing?** → Follow editing workflow

2. Creation workflow: [steps]
3. Editing workflow: [steps]

Feedback Loop Pattern

## Quality Assurance

1. Complete the task
2. Validate output against requirements
3. If validation fails:
   - Note specific issues
   - Fix problems
   - Validate again
4. Only proceed when validation passes

Anti-Patterns to Avoid

• Too verbose: Don't explain what Claude already knows
• Too many options: Provide a default with escape hatch
• Time-sensitive info: Use "old patterns" sections instead
• Inconsistent terminology: Pick one term and stick with it
• Vague descriptions: Always include specific trigger terms
• Windows paths: Always use forward slashes (even on Windows)
• Deep nesting: Keep references one level from SKILL.md
• Missing TOC: Add table of contents to long reference files

Quality Checklist

Before finalizing a skill, verify:

Core Quality:

• Name follows naming conventions (gerund form, lowercase, hyphens)
• Description is specific with trigger terms
• Description includes both what and when
• SKILL.md body is under 500 lines
• No time-sensitive information
• Consistent terminology throughout
• Examples are concrete
• File references are one level deep
• Workflows have clear steps

Structure:

• Proper YAML frontmatter
• Clear section organization
• Progressive disclosure used appropriately
• Supporting files in logical directories

Testing:

• Tested with example queries
• Skill activates on expected triggers
• Instructions are clear and complete

Supporting Files Organization

skill-name/
├── SKILL.md              # Core instructions
├── reference/            # Detailed documentation
│   ├── api.md
│   └── examples.md
├── scripts/              # Utility scripts
│   ├── helper.py
│   └── validate.py
└── templates/            # Templates and assets
    └── template.txt

Creation Example

User Request: "Create a skill for analyzing CSV files"

Questions to Ask:

• What type of analysis? (statistics, validation, transformation?)
• What triggers should activate it? (CSV files, data analysis, spreadsheets?)
• Need visualization? (charts, graphs?)
• Output format? (report, summary, processed CSV?)

Resulting Skill Structure:

~/​.claude/​skills/​analyzing-csv-data/
├── SKILL.md
├── reference/
│   └── statistics.md
└── scripts/
    ├── analyze.py
    └── validate.py

Final Steps

After creating the skill:

1. Inform the user of the location
2. Provide test queries to activate it
3. Suggest improvements based on their use case
4. Remind about iteration: Skills improve with real usage

Remember

• Keep it concise - Claude is already smart
• Make descriptions specific with trigger terms
• Use progressive disclosure for complex skills
• Test with real queries
• Iterate based on actual usage