---
title: "Skill Development for Claude Code Plugins: review, setup & limitations · Undominated.ai"
canonical: https://undominated.ai/skills/anthropic-skill-development/
description: "Guides writing a skill that ships inside a Claude Code plugin: directory layout, a third-person trigger description, an imperative SKILL.md body and progressive disclosure into references/, examples/ and scripts/."
---

# Skill Development for Claude Code Plugins: review, setup & limitations · Undominated.ai

> Guides writing a skill that ships inside a Claude Code plugin: directory layout, a third-person trigger description, an imperative SKILL.md body and progressive disclosure into references/, examples/ and scripts/.

[← Explore all skills](/skills/)

AGENT DEVELOPMENT / Anthropic

# Skill Development for Claude Code Plugins

Guides writing a skill that ships inside a Claude Code plugin: directory layout, a third-person trigger description, an imperative SKILL.md body and progressive disclosure into references/, examples/ and scripts/.

 [See setup guidance ↓](#setup)[Original source ↗](https://github.com/anthropics/claude-plugins-official/tree/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/skills/skill-development)

SOURCE REVIEW

 Reviewed 2026-10-21
 Evidence 5 linked sources
 Publisher Anthropic
 Licence [Apache-2.0 ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/LICENSE)
 Revision 41f32e407b45
 [Read what was—and wasn’t—checked ↓](#review)

“This skill should be used when the user wants to "create a skill", "add a skill to plugin", "write a new skill", "improve skill description", "organize skill content", or needs guidance on skill structure, progressive disclosure, or skill development best practices for Claude Code plugins.”

 [Anthropic · upstream description ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/skills/skill-development/SKILL.md) Our analysis follows below.

01 / THE REASONING

## Why this made the selection.

 - It adapts the skill-creator method to plugin skills, which is what separates it from the general skill-creator: plugin skills are created by hand under plugin-name/skills/<skill-name>/ instead of with init_skill.py, ship with the plugin rather than as separate ZIP files, and are reviewed with plugin-dev's skill-reviewer agent.
- It sets checkable writing rules: a third-person description that quotes the phrases users would say, an imperative body of about 1,500-2,000 words, detail moved to references/, and a checklist that every referenced file exists.

### A good fit for

 - Add a new skill directory to an existing Claude Code plugin's skills/ folder
- Rewrite a skill description in the third person with the exact phrases that should trigger it
- Move long sections of a SKILL.md into references/ and point to them from the body

### Weigh up before choosing

 - Last commit touching plugins/plugin-dev/skills/skill-development at revision 41f32e40: 2026-02-05 (2438937b).
- Its general method comes from skill-creator: references/skill-creator-original.md is labelled the full original skill-creator content, and that file still describes init_skill.py and package_skill.py, which this skill directory does not contain.
- references/skill-creator-original.md carries the frontmatter line "license: Complete terms in LICENSE.txt"; no LICENSE.txt is in the skill directory at this revision. The plugin's LICENSE is Apache-2.0.
- The skill-reviewer agent it recommends is a plugin-dev agent, outside this skill's directory, so it is present only when the whole plugin is installed.
- skill-development is distributed in plugin-dev, and that plugin's README states "MIT License - See repository for details" while plugins/plugin-dev/LICENSE is the Apache License, Version 2.0. This record follows the LICENSE file.
- The printed plugin-dev install line uses the marketplace name claude-code-marketplace, which differs from this repository's marketplace name, claude-plugins-official.

02 / THE REVIEW RECORD

## What we actually inspected.

Source review has boundaries. A clear record is more useful than a “safe” badge.

### Material inspected

 - plugins/plugin-dev/skills/skill-development/SKILL.md
- plugins/plugin-dev/skills/skill-development/references/skill-creator-original.md (frontmatter and script references)
- plugins/plugin-dev/README.md
- plugins/plugin-dev/LICENSE
- .claude-plugin/marketplace.json (name and plugin-dev entry)

### Our findings

 - The skill directory holds SKILL.md and one reference file; it ships no script and its frontmatter declares no allowed-tools.
- Step 3 creates the skill with mkdir -p plugin-name/skills/skill-name/{references,examples,scripts} and touch plugin-name/skills/skill-name/SKILL.md.
- Progressive disclosure is described as three levels: name and description always in context, the SKILL.md body when the skill triggers, bundled resources as needed.
- Plugin skills are tested by loading the plugin locally with cc --plugin-dir /path/to/plugin and asking questions that should trigger the skill.

### Not established by this review

 - No skill was created or reviewed with the skill-reviewer agent.
- references/skill-creator-original.md was read only for its frontmatter and its script references, not in full.
- Not executed: it was not installed or run.

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

## Add a skill to your workflow.

[Upstream setup instructions ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/README.md)
 - The plugin-dev README's install line is `/plugin install plugin-dev@claude-code-marketplace`. claude-code-marketplace is not this repository's marketplace, which .claude-plugin/marketplace.json names claude-plugins-official, and the line installs the entire plugin, including the skill-reviewer agent this skill relies on, so it is not a command for skill-development alone and install.command is null.
- The SKILL.md and the README both show `cc --plugin-dir` for loading a plugin from a local directory; the README prints it as `cc --plugin-dir /path/to/plugin-dev`.
- Neither line was run here.

### Before you start

 - Claude Code, with a plugin to add the skill to

### Compatibility

Claude Code

### Markdown guidance; following it creates files in a plugin's skills/ directory

 - skill-development lists no allowed-tools and bundles no script
- Its creation step runs mkdir -p and touch inside the plugin directory, and editing writes SKILL.md and any references/, examples/ or scripts/ files

### Cost model

Apache-2.0 per plugins/plugin-dev/LICENSE; the guidance does not call an outside service

04 / FOLLOW THE EVIDENCE

## The source trail.

Our notes are separate from the original resource. Check upstream before adopting a new version.

 - [skill-development SKILL.md ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/skills/skill-development/SKILL.md) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/skills/skill-development/SKILL.md Supports: upstreamDescription, summary, whySelected[0], whySelected[1], bestFor[0], bestFor[1], bestFor[2], limitations[1], limitations[3], compatibility, review.findings[0], review.findings[1], review.findings[2], review.findings[3], access.permissions[0], access.permissions[1], install.steps[1], install.prerequisites[0]
- [skill-development references/skill-creator-original.md ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/skills/skill-development/references/skill-creator-original.md) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/skills/skill-development/references/skill-creator-original.md Supports: limitations[1], limitations[2]
- [plugin-dev README ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/README.md) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/README.md Supports: install.sourceUrl, install.steps[0], install.steps[1], limitations[3], limitations[4], limitations[5], compatibility
- [plugin-dev LICENSE ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/LICENSE) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/plugin-dev/LICENSE Supports: license, limitations[2], limitations[4], access.cost
- [Marketplace file, name and plugin-dev entry ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/.claude-plugin/marketplace.json) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/.claude-plugin/marketplace.json Supports: limitations[5], install.steps[0]

KEEP COMPARING

## Other approaches to consider.

Related by category or shared topics. These are alternatives to inspect, not a measured quality order.

 [### Anthropic Skill Creator ↗ Develops skills from realistic task fixtures and baseline comparisons, with separate trigger-description experiments and human output review.](/skills/anthropics-skill-creator/)[### Command Development for Claude Code ↗ Explains how to write Claude Code slash-command files: frontmatter fields, $ARGUMENTS and positional arguments, @file references, inline Bash and plugin command patterns.](/skills/anthropic-command-development/)[### Agent Development for Claude Code Plugins ↗ Explains how to write a Claude Code plugin agent file (frontmatter, trigger description, system prompt) and ships a Bash script that checks one agent file.](/skills/anthropic-agent-development/)

 [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/)

## Continue your investigation

 - [Choose a toolkit](/tools/)
- [Inspect execution hosts](/agents/)
- [Check tool access](/mcp-servers/)
- [Check project dependencies](/check-repo/)
