Purpose

Creates new skills following the agent skills format: YAML frontmatter for routing metadata plus a markdown body with Purpose, When to use, Operating procedure, Output defaults, and Failure handling. The description field is routing logic—it determines whether the skill fires.

When to use this skill

Use when:

• User says "create a skill for X", "write a skill that...", "I need a skill to handle..."
• Repeated task pattern needs to be captured as reusable procedure
• Capability should be packaged for distribution or reuse across projects
• New agent specialization needed for a repo's agent team

Do NOT use when:

• Skill exists and needs improvement (use

  skill-refinement

  or

  skill-improver

  )
• Only description/trigger needs fixing (use

  skill-trigger-optimization

  )
• User wants a one-off prompt, not a reusable skill
• Skill needs adaptation to different context (use

  skill-adaptation

  )

Operating procedure

Step 1 — Define the skill's job in one sentence

"This skill [verb] when [trigger] and produces [output]."

If you cannot write this sentence, the scope is wrong—narrow until it works.

Step 2 — Choose the name

• Lowercase, hyphens, 2-4 words
• Name describes what it does (verb-noun), not when it's used
• Under 64 characters
• Examples:

  premortem

  ,

  gap-analysis

  ,

  acceptance-criteria-hardening

Step 3 — Write the YAML frontmatter

---
name: skill-name
description: >-
  [Main action verb] [specific object] when [task conditions].
  Use this for [2-3 realistic trigger phrases in quotes].
  Do not use for [adjacent non-matching cases with named alternatives].
license: Apache-2.0
compatibility:
  clients: [openai-codex, gemini-cli, opencode, github-copilot]
metadata:
  owner: [owner-name]
  domain: [domain-name]
  maturity: draft|stable|deprecated
  risk: low|medium|high
  tags: [relevant, tags]
---

Description field rules (THIS IS ROUTING LOGIC)

The description determines whether the skill fires. It is the highest-leverage field.

A strong description must:

1. Start with the main action verb (not a noun phrase)
2. Include 2-3 realistic trigger phrases users would actually say
3. State what the skill produces
4. End with "Do not use for..." naming adjacent skills

Weak: "Helps with PDFs." Weak: "A skill for managing deployments." Strong: "Extract structured data from PDFs when the task involves tables, scanned pages, or layout-dependent interpretation. Use this for requests like 'extract the table from this PDF', 'convert this PDF to CSV', or 'parse the scanned invoice'. Do not use for plain-text summarization when PDF text is already clean (use text-summarization)."

Lint checks — flag a description if it:

• Is under 12 words
• Has no action verb as first word
• Has no "when" or equivalent condition
• Lacks concrete trigger examples
• Is obviously generic or could apply to multiple skills
• Has no negative boundary

Frontmatter fields

name

description

license

compatibility.clients

metadata.owner

metadata.domain

metadata.maturity

metadata.risk

metadata.tags

Step 4 — Write the body sections

Every SKILL.md body must contain these sections in order:

Section: Purpose (required)

2-3 sentences. What problem does this solve? What output does it produce? No filler.

Section: When to use this skill (required)

• "Use when:" — 4-6 specific trigger phrases or observable conditions
• "Do NOT use when:" — 3-4 confusion cases with named alternatives

Section: Operating procedure (required)

Numbered steps, each starting with a verb, each completable and verifiable.

Rules:

• No meta-commentary like "Keep scope explicit" or "Think about the approach"
• No hedge verbs: "Consider", "You might want to", "Ensure"
• Use action verbs: "Read", "List", "Write", "Check", "Run", "Compare"
• Each step must be independently executable

Section: Output defaults (required)

Exact format with template showing specific section names, field names, or schemas.

Bad: "Structured markdown with clear next steps" Good: Template with named sections, tables, or code blocks

Section: References (recommended)

Real URLs to authoritative documentation. Not vague "see docs".

Section: Failure handling (required)

Name the 2-3 most common failure modes with specific recovery actions.

Bad: "If something goes wrong, report the issue" Good: "If target file does not exist: stop, report missing path, ask user to confirm location"

Step 5 — Validate against anti-patterns

Check for:

• No circular triggers ("when task involves X")
• No boilerplate procedure steps
• No generic output defaults
• No vague failure handling
• Description starts with action verb
• "Do NOT use when" section present with alternatives named

Step 6 — Create the skill folder

skill-name/
├── SKILL.md          # Frontmatter + body sections
├── scripts/          # Optional: deterministic automation
├── references/       # Optional: large docs for progressive disclosure
└── evals/            # Optional: trigger and output tests

Only create scripts/, references/, evals/ directories if they contain actual files. Do not create empty directories.

Output defaults

A complete SKILL.md file following the template above, inside a properly named folder. If the skill is complex enough, include references/ or scripts/ with content.

References

• https://docs.github.com/en/copilot/concepts/agents/about-agent-skills — GitHub agent skills overview
• https://docs.github.com/en/copilot/how-tos/use-copilot-agents/coding-agent/create-skills — GitHub skill creation guide
• https://developers.openai.com/codex/skills — OpenAI Codex skill format
• https://docs.anthropic.com/en/docs/claude-code/overview — Anthropic Claude Code
• https://geminicli.com/docs/cli/skills/ — Gemini CLI skills

Failure handling

• Scope too broad: If skill handles multiple distinct tasks, split into separate skills via

  skill-variant-splitting

• Cannot write one-sentence definition: Scope is wrong—narrow until it works
• Overlaps existing skill: Check catalog; merge or explicitly differentiate in description
• No clear trigger phrases: Ask user what words they'd use when they need this capability
• Description passes lint but skill never fires: Add more trigger phrase variations; test with

  skill-testing-harness