Plugin Ecosystem

Overview

Plugins bundle Skills, Agents, Hooks, MCP servers, and LSP servers into distributable extension packages. Discovered and installed through Plugin Marketplaces.

Plugin Directory Structure

my-plugin/
├── .claude-plugin/
│   └── plugin.json        # Plugin manifest (required)
├── skills/                # Agent Skills (SKILL.md files)
│   └── my-skill/
│       └── SKILL.md
├── commands/              # Skills as Markdown files
│   └── hello.md
├── agents/                # Custom agent definitions
│   └── reviewer.md
├── hooks/                 # Event handlers
│   └── hooks.json
├── .mcp.json              # MCP server configurations
├── .lsp.json              # LSP server configurations
└── settings.json          # Default settings

Plugin Manifest

{
  "name": "my-plugin",
  "description": "What the plugin does",
  "version": "1.0.0",
  "author": { "name": "Author Name" }
}

Skill Namespacing

Plugin skills use plugin-name:skill-name format (e.g., /my-plugin:hello), avoiding conflicts with project-level skills.

Standalone vs Plugin

Testing Plugins

# Load local plugin directory for testing
claude --plugin-dir ./my-plugin

# Reload in session (no restart needed)
/reload-plugins

Plugin Security Restrictions

Plugin subagents do NOT support these frontmatter fields (ignored when loading from a plugin):

• hooks
• mcpServers
• permissionMode

This ensures plugins cannot bypass security restrictions.

Official Plugins (Anthropic Marketplace)

Code Intelligence (LSP Plugins)

Code intelligence support for 11 languages:

External Integration Plugins

Development Workflow Plugins

Output Style Plugins

Create Your Own Marketplace

You can create and publish your own plugin catalog for teams or communities. Plugins support auto-update systems.