Generating Skills
Creates Claude Code agent skill files from expertise descriptions. Use when someone describes their professional methodology or workflow that should be captured as a reusable skill.
YAML--- name: analyzing-data description: Analyzes datasets to identify patterns and insights. Use when exploring new data or validating hypotheses. ---
Load your dataset and run initial exploration:
Pythonimport pandas as pd df = pd.read_csv('data.csv') print(df.info()) print(df.describe())
- Load and inspect data structure
- Check for missing values and outliers
- Generate summary statistics
- Create visualizations for key variables
- Document findings and recommendations
- Parse Input - Extract the core expertise from the description
- Structure Content - Organize into the 5 required sections
- Create YAML Header - Generate gerund-form name and trigger-based description
- Write Quick Start - Provide immediate actionable example
- Document Workflow - Break down the process into clear steps with checklist
- Add Examples - Include concrete input/output pairs
- List Best Practices - Capture key guidelines
- Identify Pitfalls - Note common mistakes to avoid
Progress:
- Extract methodology from input
- Create YAML frontmatter with gerund name
- Write Quick Start section with code/steps
- Document step-by-step workflow
- Add concrete examples
- List best practices
- Identify common pitfalls
- Review for conciseness (~500 lines max)
Example 1:
Input: "I'm a data scientist who specializes in exploratory data analysis. I always start by loading the data, checking data types, looking for missing values, then creating visualizations."
Output: A skill named analyzing-data with Python code examples for pandas operations, visualization steps, and a systematic workflow checklist.
Example 2:
Input: "I help companies write better proposals by structuring them with executive summary, problem statement, solution approach, timeline, and budget."
Output: A skill named writing-proposals with template structure, example sections, and review criteria.
- Use gerund form for skill names (verb-ing format)
- Include trigger phrases in description ("Use when...")
- Start with actionable Quick Start, no preamble
- Provide concrete code examples or step templates
- Keep explanations concise - assume Claude knows basics
- Include progress checklists for complex workflows
- Focus on specific methodology, not general advice
- Aim for ~500 lines maximum length
- Don't use first person ("I do...") or second person ("You should...")
- Don't explain basic concepts Claude already knows
- Don't create multiple options - provide the recommended approach
- Don't ask clarifying questions if input is unclear - create minimal placeholder
- Don't exceed 64 characters for skill names
- Don't forget YAML frontmatter at the very beginning
- Don't include preamble before the skill content