claude-code-meta-skill/META-SKILL-GENERATOR-GUIDE.md
2026-05-21 10:23:18 +00:00

7.6 KiB

Meta Skill Generator - User Guide

What Is This?

The Meta Skill Generator is a Claude Code skill that creates other Claude Code skills. It intelligently analyzes your requirements and separates operations into:

  1. Go scripts - For deterministic, performance-critical operations
  2. Python scripts - For operations requiring specialized libraries
  3. Agent workflows - For tasks requiring reasoning and context

Key Features

Workflow analysis - Automatically identifies which operations should be scripts vs workflows Go script generation - Creates efficient, production-ready Go binaries Best practices - Follows skill-creator patterns and conventions End-to-end workflow - From requirements gathering to packaged .skill file Comprehensive references - Go patterns, workflow analysis, and examples

Quick Start

1. Interactive Skill Creation

scripts/init_skill_with_analysis.py my-new-skill --path ./skills

This will guide you through:

  • Gathering example use cases
  • Analyzing which operations should be scripts
  • Specifying Go script details
  • Generating the complete skill structure

2. Generate Individual Go Scripts

scripts/generate_go_script.py \
  --name pdf-to-images \
  --description "Convert PDF pages to PNG images" \
  --input "PDF file path" \
  --output "Directory of PNG files" \
  --logic "Extract each page as separate image at 300 DPI" \
  --skill-path ./my-skill

This creates:

  • Complete Go source code with error handling
  • Build script for compilation
  • Documentation in SKILL.md

3. Analyze Existing Requirements

# Save your requirements to a file
cat > requirements.txt << EOF
Extract text from PDF files
Convert CSV to JSON format
Analyze sentiment in documents
Rotate images by 90 degrees
EOF

# Run analysis
scripts/analyze_workflow.py --examples requirements.txt

This shows which operations should be Go scripts, Python scripts, or agent workflows.

How It Works

Workflow Analysis Algorithm

The analyzer looks for keywords to categorize operations:

Deterministic keywords → Go/Python scripts:

  • convert, transform, parse, extract, validate
  • format, encode, decode, compress, resize
  • sort, filter, calculate, merge, split

Dynamic keywords → Agent workflows:

  • analyze, understand, interpret, decide
  • suggest, recommend, summarize, explain
  • evaluate, assess, identify, classify

Performance indicators → Go preferred:

  • large file, batch, thousands, concurrent
  • parallel, binary, low-level, stream

Library indicators → Python preferred:

  • pandas, numpy, machine learning, plot
  • api client, visualization

Go Script Template

Generated scripts include:

  • CLI parsing with --help and --verbose flags
  • Input validation before processing
  • Error handling with descriptive messages
  • Progress reporting for long operations
  • Proper exit codes (0=success, 1=error, 2=invalid input)
  • Helper functions based on operation type

Skill Structure

my-skill/
├── SKILL.md              # Main skill documentation
├── scripts/
│   ├── script1.go        # Generated Go source
│   ├── script2.go
│   ├── build_script1.sh  # Build scripts
│   ├── build_script2.sh
│   └── bin/              # Compiled binaries
│       ├── script1
│       └── script2
├── references/           # Detailed documentation
│   └── advanced-guide.md
└── assets/               # Templates, images, etc.
    └── template.txt

Examples

Example 1: PDF Processing Skill

Requirements:

  • Extract text from PDFs
  • Convert PDFs to images
  • Merge multiple PDFs
  • Summarize PDF content

Analysis Result:

  • pdf-extract-text → Go script (deterministic, parsing)
  • pdf-to-images → Go script (deterministic, binary)
  • merge-pdfs → Go script (deterministic, file ops)
  • Summarize → Agent workflow (requires understanding)

Example 2: Data Analytics Skill

Requirements:

  • Convert CSV to JSON
  • Validate data against schema
  • Plot trends
  • Find insights

Analysis Result:

  • csv-to-json → Go script (simple conversion)
  • validate-schema → Go script (deterministic rules)
  • plot-data → Python script (needs matplotlib)
  • Find insights → Agent workflow (interpretation)

Best Practices

When to Create a Skill

Create a skill when:

  • You're doing the same operations repeatedly
  • Operations have clear, reusable patterns
  • You want consistent behavior across tasks
  • Performance matters (batch processing, large files)

Don't create a skill for:

  • One-off tasks
  • Highly variable, context-dependent operations
  • Tasks that change frequently

Go vs Python vs Agent

Use Go for:

  • File format conversions
  • Batch processing (1000s of items)
  • Binary/low-level operations
  • Performance-critical paths

Use Python for:

  • Data science (pandas, numpy)
  • ML/AI tasks
  • Complex API clients
  • Visualization

Use Agent workflows for:

  • Content analysis
  • Creative tasks
  • Decision-making
  • Context-dependent branching

Skill Design Tips

  1. Start with examples - Gather 3-5 concrete use cases
  2. Identify patterns - Look for repeated code or workflows
  3. Separate concerns - Scripts for deterministic, workflows for reasoning
  4. Keep SKILL.md lean - Move details to references
  5. Test everything - Run scripts before packaging
  6. Document clearly - Include examples for every operation

Reference Documentation

Included References

  • go-patterns.md - Battle-tested Go patterns for common operations

    • File processing (streaming, line-by-line)
    • Concurrent processing (worker pools, rate limiting)
    • Progress reporting
    • Error handling
    • CLI patterns
    • Data processing (CSV, JSON)
  • workflow-analysis.md - Deep dive on deterministic vs dynamic

    • Decision framework
    • Real-world examples
    • Common patterns
    • Anti-patterns to avoid
  • skill-examples.md - Complete examples of well-designed skills

    • PDF Tools
    • Data Processor
    • Image Tools
    • Pattern analysis

Troubleshooting

"Go compiler not found"

Install Go: https://go.dev/doc/install

Or skip Go generation:

scripts/init_skill_with_analysis.py my-skill --path . --skip-go

"Script compilation failed"

Check the error message. Common issues:

  • Missing imports (add to imports in template)
  • Undefined functions (implement TODOs)
  • Syntax errors (review generated code)

"Skill validation failed"

Common validation errors:

  • Missing description in frontmatter
  • Description too short (<50 chars)
  • SKILL.md not found
  • Invalid skill name (use hyphen-case)

Advanced Usage

Customize Go Template

Edit generate_go_script.py to modify:

  • Template structure
  • Import inference logic
  • Helper function generation
  • CLI flag patterns

Add Custom Analysis Rules

Edit analyze_workflow.py to:

  • Add domain-specific keywords
  • Adjust scoring thresholds
  • Create custom operation types

Extend References

Add your own patterns:

cd my-skill/references
echo "# My Patterns" > custom-patterns.md
# Add your documentation

Reference from SKILL.md:

For advanced patterns, see [references/custom-patterns.md](references/custom-patterns.md)

Support

This skill follows the skill-creator guidelines. For questions about:

  • Skill structure → See skill-creator SKILL.md
  • Go patterns → See references/go-patterns.md
  • Workflow analysis → See references/workflow-analysis.md
  • Examples → See references/skill-examples.md

License

This skill is created using the skill-creator framework and includes example patterns from the Claude Code skill ecosystem.