Skill Creator

Creates well-structured Skills following Anthropic's official best practices.

When to Create a Skill

Good candidates:

• Repeated 3+ times
• Complex multi-step workflows
• Error-prone processes
• Domain-specific knowledge

Not worth codifying:

• One-off tasks
• Highly variable processes
• Trivial single-step actions

SKILL.md Structure

Required Frontmatter

---
name: processing-documents
description: "Processes documents and extracts data. Use when working with PDF or Word files. Triggers on: process document, extract data."
---

Field requirements:

Naming Convention

Use gerund form (verb + -ing):

• processing-pdfs
• analyzing-spreadsheets
• managing-databases

Avoid: helper, utils, tools, anthropic-*, claude-*

Writing Descriptions

Always use third person (injected into system prompt):

# Good
description: "Processes Excel files and generates reports"

# Avoid
description: "I can help you process Excel files"
description: "You can use this to process Excel files"

Be specific and include triggers:

description: "Analyzes BigQuery data and generates reports. Use when querying sales metrics or creating dashboards. Triggers on: analyze data, bigquery report, sales metrics."

Skill Body Template

# [Skill Title]

[One sentence describing what this skill does]

## Quick Start

[Minimal example to get started - under 50 tokens]

## Process

1. [Step 1]
2. [Step 2]
3. [Step 3]

## Advanced Features

**[Feature A]**: See [FEATURE_A.md](FEATURE_A.md)
**[Feature B]**: See [FEATURE_B.md](FEATURE_B.md)

Best Practices

Be Concise

Only add context Claude doesn't already have. Challenge each piece:

• "Does Claude really need this explanation?"
• "Can I assume Claude knows this?"

Good (~50 tokens):

## Extract PDF text

Use pdfplumber:
\`\`\`python
import pdfplumber
with pdfplumber.open("file.pdf") as pdf:
    text = pdf.pages[0].extract_text()
\`\`\`

Bad (~150 tokens): Explaining what PDFs are and how libraries work.

Keep Under 500 Lines

If content exceeds 500 lines, split into separate files using progressive disclosure.

One-Level Deep References

Keep all reference files linked directly from SKILL.md:

# SKILL.md
**Basic usage**: [instructions here]
**Advanced features**: See [advanced.md](advanced.md)
**API reference**: See [reference.md](reference.md)

Avoid Time-Sensitive Info

Use "old patterns" section instead of dates:

## Current method
Use the v2 API endpoint.

## Old patterns
<details>
<summary>Legacy v1 API (deprecated)</summary>
The v1 API used different endpoints...
</details>

Use Consistent Terminology

Choose one term and stick with it:

• Always "API endpoint" (not mix with "URL", "route", "path")
• Always "extract" (not mix with "pull", "get", "retrieve")

Directory Structure

skill-name/
├── SKILL.md              # Main instructions (< 500 lines)
├── ADVANCED.md           # Advanced features (loaded as needed)
├── REFERENCE.md          # API reference (loaded as needed)
└── scripts/
    └── utility.py        # Executed, not loaded into context

Checklist

Before saving:

• Name uses gerund form and kebab-case
• Description is third person and specific
• Description includes what it does AND when to use it
• SKILL.md body under 500 lines
• No unnecessary explanations Claude already knows
• References are one level deep
• No time-sensitive information
• Consistent terminology throughout