What does this skill do, and when should you use it?
This is the skill-development skill from the plugin-dev plugin in anthropics/claude-plugins-official. It guides authors who need to create or improve skills for Claude Code plugins. The guidance covers skill directories, YAML frontmatter, trigger descriptions, progressive disclosure, and the organization of scripts, references, examples, and assets. It also defines a workflow from use-case analysis through structure creation, editing, validation, testing, and iteration. The material is specifically oriented toward Claude Code plugin development.
Guides authors in creating or editing a skill directory containing SKILL.md; defining name and description frontmatter; planning scripts, references, examples, and assets; keeping the main body lean at roughly 1,500–2,000 words; using third-person trigger descriptions and imperative or infinitive instructions; checking structure, trigger phrases, writing style, referenced files, examples, and scripts; testing a local plugin with cc --plugin-dir /path/to/plugin; and iterating after real use. It does not claim to automatically run validation scripts, call MCP services, or generate files.
- A developer adding a skill to a Claude Code plugin can use the workflow to identify use cases, reusable resources, and directory structure.
- A skill author improving activation behavior can rewrite the description with concrete phrases users are likely to say.
- A maintainer dealing with an oversized skill can move detailed patterns, API material, and edge cases into references/.
- A plugin developer checking whether a skill loads correctly can validate its frontmatter and structure and test locally with `cc --plugin-dir`.
How do you install this skill?
- No commands or tests were executed; the referenced references/, examples/, scripts/, and their runnability were not individually verified in this evidence set.
- The skill guides users to create executable scripts and invoke external tools but does not define permission boundaries, user confirmation, sensitive-data handling, or rollback procedures.
- Chinese-language behavior and mainland-China network compatibility are not specified; reachability must be checked separately if the workflow depends on overseas documentation or services.
- The README warns that plugins may include MCP servers, files, or software not controlled or verified by Anthropic; official provenance does not guarantee this skill's execution safety or effectiveness.
- Shell / CLI
- Local filesystem
The skill is part of the anthropics/claude-plugins-official plugin collection. The README says plugins can be installed in Claude Code with /plugin install {plugin-name}@claude-plugins-official or discovered through /plugin > Discover; the supplied source does not state the exact marketplace name for plugin-dev. The repository is licensed under Apache-2.0.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/anthropics/claude-plugins-official.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/plugins/plugin-dev/skills/skill-development" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- create a skill,
- add a skill to plugin,
- improve skill description,
- organize skill content.
In Claude Code, make a concrete request such as “create a skill,” “add a skill to plugin,” “improve skill description,” or “organize skill content.” For local plugin development, run cc --plugin-dir /path/to/plugin, then ask questions that should trigger the skill and verify that it loads correctly.
What are this skill's strengths and limitations?
- Provides an end-to-end workflow from use-case analysis through testing and iteration.
- Clearly specifies frontmatter, trigger phrases, imperative writing, and progressive disclosure practices.
- Covers common plugin-skill resources including scripts, references, examples, and assets.
- Includes structure, content-quality, progressive-disclosure, and testing checklists.
- The guidance is centered on Claude Code plugins, so other Agent Skills clients would require changes to packaging and test procedures.
- The supplied source does not establish that this skill includes an executable automated validator or a complete test suite.
- The README does not provide plugin-dev's exact marketplace installation identifier.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Claude Code Plugin Skill Development this page ✓ Anthropic · Official | 51 · Use with care | ★ 38k | 11d ago | Apache-2.0 |
| Claude Code Skill Development ✓ Anthropic · Official | 46 · Use with care | ★ 151k | 11d ago | — |
| Agent Skill Blueprint | 40 · Not recommended | ★ 18k | 13d ago | MIT |
| Create New Skills (Meta-Skill) | 62 · Recommended | ★ 1.5k | 8mo ago | — |
| Claude Code Agent Development Guide ✓ Anthropic · Official | 54 · Use with care | ★ 38k | 11d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill is primarily documentation and directory guidance; no malware, credential theft, covert exfiltration, or destructive default is evident. However, it recommends creating and testing executable scripts without specifying least privilege, user confirmation, sensitive-data handling, dependency security, external effects, or rollback, so points are deducted.
The process is structured and includes checks for structure, references, scripts, and trigger phrases, with some diagnosable failure criteria. However, it provides no executable validation implementation, and the existence or runnability of referenced resources and examples cannot be confirmed from this file; static review cannot establish key-path reproduction, so the score remains below 10.
The frontmatter supplies reasonably specific trigger phrases, and the body covers skill creation, improvement, organization, and testing scenarios. It does not clearly define non-fit boundaries, input/output contracts, false-trigger boundaries, compatibility expectations, or Chinese-language support, so environment and invocation fit receive deductions.
Information architecture, progressive disclosure, directory conventions, naming guidance, version metadata, resource references, examples, limitations, and testing guidance are comparatively strong. The skill lacks its own maintainer, changelog, update path, and complete installation/dependency notes, while some quality claims depend on other files not assessed here, so points are deducted.
The skill offers an actionable creation workflow, directory templates, trigger-description rules, and a validation checklist that plausibly covers the core task. No representative outputs, real usage feedback, or statically verifiable success cases are provided, and implementation plus review remain necessary, so the static ceiling applies.
The assessed SKILL.md and its referenced original source are auditable primary materials with explicit checklists. There is no independent reproduction record, committed test suite covering the skill's key paths, or third-party corroboration; claims about excellent examples and effectiveness are largely author assertions, so the score does not exceed the static-review ceiling.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →