Create Skill

Create Skill

This skill helps you scaffold new agent skills that conform to the Agent Skills specification and the dotnet/skills repository conventions.

When to Use

• Creating a new skill from scratch
• Generating a SKILL.md file with proper frontmatter
• Setting up the skill directory structure with optional folders
• Ensuring compliance with agentskills.io specification

When Not to Use

• Modifying existing skills (edit directly instead)
• Diagnosing or fixing a skill that fails its evaluation (use improve-skill-quality)
• Writing the skill's eval.yaml (use create-skill-test)
• Creating custom agents (use the agents/ directory pattern)

Inputs

Workflow

Step 1: Validate the skill name

Ensure the name:

• Contains only lowercase letters, numbers, and hyphens
• Does not start or end with a hyphen
• Does not contain consecutive hyphens
• Is between 1-64 characters

Step 2: Write the description — it is the router

The description is the only text the runtime sees when deciding whether to load the skill. A perfect body behind a weak description never runs.

---
name: <skill-name>
description: <what it does>. USE FOR: <symptoms, error codes, artifact names, quoted user requests>. DO NOT USE FOR: <nearby-but-wrong intents, with the skill that owns them>.
---

• Lead with an action verb and use the user's own words: symptoms, error codes (CS1501, MSTEST0014), artifact names (.testsettings, binlog), and requests phrased as a developer would type them.
• Partition against sibling skills on the real discriminator, not the topic. "Does the abstraction already exist?" separates two skills; "testing" does not. Add the matching exclusion to both siblings.
• Claim the ambiguous words that would otherwise route to a sibling. If prompts say "review my tests" and a sibling owns "review", say so explicitly.
• Check every DO NOT USE FOR clause against the scenarios the skill exists to serve — an exclusion like "already on v3" can lock out the post-upgrade fixes that are the skill's purpose.
• Budget: 1,024 characters per description, and the whole plugin's rendered skill menu is also budgeted. A helper skill users should never invoke directly can set disable-model-invocation: true to free menu space while staying invocable by name.
• No XML tags. Claude (claude.ai and Claude Code marketplace sync) rejects descriptions containing tag-like text such as Vector<T> or <Import>. Spell it out in words ("generic Vector type", "Import element"); skill-validator check fails on it.

Step 3: Write for delta over the baseline model

Every skill is scored head-to-head against the same model with no skill loaded. Content the model already produces unaided is worth zero; content that makes it slower or more hedged is worth less than zero. See improve-skill-quality/references/writing-for-baseline-delta.md for the full evidence.

Do not over-correct: a skilled answer shorter and less actionable than the baseline's still loses.

Step 4: Create the skill directory

plugins/<plugin>/skills/<skill-name>/
└── SKILL.md

Step 5: Generate SKILL.md with frontmatter

Create the file with the frontmatter drafted in Step 2.

Step 6: Add body content sections

Include these recommended sections:

1. Purpose: One paragraph describing the outcome
2. When to Use: Bullet list of appropriate scenarios
3. When Not to Use: Boundaries and exclusions
4. Inputs: Table of required and optional inputs
5. Workflow: Numbered steps with checkpoints
6. Validation: How to confirm the skill worked correctly
7. Common Pitfalls: Known traps and how to avoid them

Step 7: Add optional directories (if needed)

plugins/<plugin>/skills/<skill-name>/
├── SKILL.md
├── scripts/       # Executable code agents can run
├── references/    # Additional documentation loaded on demand
└── assets/        # Templates, images, data files

Reference bundled files with paths relative to the directory that contains SKILL.md, such as references/details.md or scripts/validate.ps1. Do not tell the agent to search for the skill's installation directory. If a missing bundled file reduces the result quality, allow one listing of the known bundled-file directory and require the agent to report the reduced coverage.