Skill Authoring

Create skills that follow the Agent Skills specification—an open format supported by Claude Code, Cursor, VS Code, GitHub, and other agent products.

Quick Start

1. Clarify requirements

   • What capability does this provide?
   • What triggers should activate it?
   • Location: personal (~/.claude/skills/) or project (.claude/skills/)?

2. Choose template from templates/:

   • simple-skill.md - Single SKILL.md file
   • tool-restricted-skill.md - Limited tool access
   • multi-file-skill.md - With supporting files
   • script-based-skill.md - Includes executable scripts

3. Generate name

   • Pattern: lowercase, hyphens only (e.g., pdf-processing, code-review)
   • Gerund form recommended: processing-pdfs, analyzing-data
   • Must match directory name

4. Write SKILL.md

   • Valid YAML frontmatter
   • Third-person description with triggers
   • Clear, concise instructions
   • Concrete examples

5. Add supporting files (if needed)

   • references/ - detailed documentation
   • scripts/ - executable utilities
   • assets/ - templates, data files

6. Validate

   • Use skills-ref validate ./my-skill or skills-check

Directory Structure

skill-name/
├── SKILL.md           # Required: instructions + metadata
├── scripts/           # Optional: executable code
├── references/        # Optional: documentation
└── assets/            # Optional: templates, resources

SKILL.md Format

---
name: skill-name
description: What it does and when to use it. Include trigger keywords.
license: Apache-2.0                    # optional
compatibility: Requires git and jq     # optional
metadata:                              # optional
  author: your-org
  version: "1.0"
allowed-tools: Read Grep Glob          # optional, experimental
---

# Skill Name

## Quick Start
[Minimal working example]

## Instructions
[Step-by-step guidance]

## Examples
[Concrete, runnable examples]

For cross-tool compatibility details, see the [references/](references/) directory.

Frontmatter Fields

Refer to the Agent Skills Specification for complete schema.

Core Principles

Concise is key

The context window is shared. Only include what the agent doesn't already know. Challenge each paragraph—does it justify its token cost?

Third-person descriptions

Descriptions inject into system prompt. Use third person:

• ✓ "Extracts text from PDFs"
• ✗ "I can help you extract text from PDFs"

Progressive disclosure

Keep SKILL.md under 500 lines. Move details to:

• references/ - Detailed docs, API references
• scripts/ - Executable utilities (code never enters context)
• assets/ - Templates, data files

Token loading:

1. Metadata (~100 tokens): name + description loaded at startup for all skills
2. Instructions (<5000 tokens): SKILL.md body loaded when skill activates
3. Resources (as needed): files loaded only when referenced

Description Formula

[WHAT] + [WHEN] + [TRIGGERS]

description: Extracts text and tables from PDF files, fills forms, merges documents. Use when working with PDF files or when the user mentions PDFs, forms, or document extraction.

Include trigger keywords that users would naturally say when they need this skill.

Naming Requirements

• Lowercase letters, numbers, hyphens only
• Cannot start/end with hyphen or contain --
• Must match parent directory name
• Cannot contain anthropic or claude

Recommended: Gerund form (processing-pdfs, reviewing-code)

Tool Restrictions

Use allowed-tools to limit capabilities (experimental):

# Read-only
allowed-tools: Read Grep Glob

# With shell access
allowed-tools: Bash(git:*) Bash(jq:*) Read

# Full access (default)
# Omit field entirely

Skill Types

Simple - Single SKILL.md file for basic capabilities

Tool-Restricted - Uses allowed-tools to limit tool access

Multi-File - SKILL.md with supporting references/ documentation

Script-Based - Includes executable scripts in scripts/

Validation

Use the official skills-ref library:

skills-ref validate ./my-skill

Or use skills-check for integrated validation and review.

Quick Self-Check

Before finalizing, verify:

• YAML frontmatter valid (opens/closes with ---)
• Name lowercase, matches directory
• Description uses third-person voice
• Description includes 3-5 trigger keywords
• SKILL.md under 500 lines
• All referenced files exist

Common Description Mistakes

• ❌ "Helps with files" → Too vague
• ❌ "I can help you..." → Wrong voice (first-person)
• ❌ "You can use this to..." → Wrong voice (second-person)
• ❌ No "Use when..." → Missing triggers

Use the skills-check skill to validate your skill before finalizing.

Best Practices

1. Concise: Challenge each paragraph—does the agent need this?
2. Third-person: Descriptions inject into system prompt
3. Under 500 lines: Keep SKILL.md focused
4. One level deep: File references should be direct from SKILL.md
5. Single responsibility: One skill, one capability
6. Concrete examples: Show real usage, not abstract concepts
7. Test with real queries: Verify the skill triggers correctly

After Creation

1. Check: Run skills-check to validate and review
2. Test: Ask a question that should trigger it
3. Iterate: Fix issues and re-check until passing
4. Share: Commit to git or distribute via plugin

Troubleshooting

# Find all skills
find ~/.claude/skills .claude/skills -name "SKILL.md" 2>/dev/null

# Check for tab characters (YAML requires spaces)
grep -P "\t" SKILL.md

# Find broken markdown links
grep -oE '\[[^]]+\]\([^)]+\)' SKILL.md | while read link; do
  file=$(echo "$link" | sed 's/.*(\(.*\))/\1/')
  [ ! -f "$file" ] && echo "Missing: $file"
done

Resources

• Agent Skills Specification
• Best Practices Guide
• Example Skills Repository
• skills-ref Validation Library