The Skills and Plugins Ecosystem

Before you build anything, you need to understand the building blocks. Cowork’s ecosystem has four components, and mixing them up is the fastest way to waste time.

What You’ll Learn

By the end of this lesson, you’ll be able to:

• Name the four components: skills, plugins, slash commands, sub-agents
• Explain when to use each one
• Navigate Anthropic’s official plugin library

How This Course Works

This is an intermediate course — 8 lessons, about 3 hours total. Each lesson builds on the previous one. By Lesson 3 you’ll be building skills, by Lesson 5 you’ll be wiring up MCP connectors, and by Lesson 8 you’ll publish a complete plugin.

You should be comfortable with Cowork basics before starting. If terms like “connectors” or “context window” are unfamiliar, take the Claude Cowork Essentials course first.

The Four Components

Skills

A skill is Claude’s domain expertise, encoded in a file. It’s a markdown file called SKILL.md that tells Claude how to do something specific.

For example, a “contract review” skill might tell Claude:

• What to look for in contracts (liability clauses, auto-renewal, termination terms)
• What format to output the review in
• What questions to flag for legal review

When you give Cowork a contract to review, it reads the skill and applies that expertise automatically.

Skills are the smallest unit. One file, one capability. No code required.

Plugins

A plugin bundles everything together. Think of it as a toolkit for a specific role or domain.

A “Marketing” plugin might include:

• 6 skills (campaign analysis, content calendar, competitive research, etc.)
• 3 slash commands (/new-campaign, /analyze-metrics, /content-brief)
• 2 MCP connectors (Google Analytics, HubSpot)
• 1 sub-agent for multi-step campaign planning

Plugins are what you install and share. Skills are what make them smart.

Slash Commands

Slash commands are explicit triggers. You type /command-name and it runs. Unlike skills, which Claude applies automatically when relevant, commands only fire when you invoke them.

Good for: repetitive actions you want to trigger on demand. /weekly-report, /clean-data, /prep-meeting.

Sub-agents

Sub-agents are specialized agents within a plugin that handle complex subtasks. If a marketing plugin needs to analyze competitor data and generate campaign ideas, those might be two different sub-agents — each with their own skills and context.

Most users don’t build sub-agents directly. They come packaged inside plugins.

✅ Quick Check: What’s the difference between a skill and a slash command? (Skills are applied automatically when relevant; commands are triggered manually by typing /name.)

How They Fit Together

Plugin
├── Skills (auto-applied domain expertise)
├── Slash Commands (manually triggered actions)
├── MCP Connectors (links to external tools)
└── Sub-agents (specialized task handlers)

Everything is file-based. Skills are markdown. The manifest is JSON. Commands are markdown with frontmatter. No build steps, no infrastructure, no deployment pipelines.

That’s the beauty of this system — a marketing manager who’s never written code can build a plugin in an afternoon.

Anthropic’s Official Plugins

Anthropic released 15 open-source plugins on GitHub (anthropics/knowledge-work-plugins). Here’s what’s available:

All open source, all free, all customizable.

When to Use What

Key Takeaways

• Skills = one capability in a SKILL.md file, auto-discovered by Claude
• Plugins = bundles of skills + commands + connectors + sub-agents
• Slash commands = manually triggered, skills = automatically applied
• Anthropic’s 15 official plugins cover major business domains with 85+ skills
• Everything is file-based — markdown and JSON, no code required

Up Next

Next lesson, you’ll install official plugins, customize them for your role, and see how Claude applies skills automatically. Time to use what other people built.