A Claude Code plugin bundles skills, agents, hooks and MCP servers into one install: /plugin install <plugin>@<marketplace>. Install it at project scope and everyone who opens the repo gets it.
A Claude Code plugin is a directory of skills, agents, hooks, MCP servers and other components that Claude Code installs and loads as one unit. Plugins usually come from a marketplace: a catalog that lists plugins and where to fetch each one. A team can install the same setup with a couple of commands instead of copying files between laptops.
What a plugin can contain
Every component is optional, and each has its own place in the plugin folder:
MCP servers in .mcp.json and LSP servers in .lsp.json.
Output styles, themes, monitors, executables in bin/ and default settings.
A manifest at .claude-plugin/plugin.json, in which only name is required.
How to install a Claude Code plugin
Run /plugin to browse the Discover, Installed and Marketplaces tabs, or use the commands directly:
/plugin marketplace add <owner/repo or URL> adds a marketplace. Anthropic's official marketplace, claude-plugins-official, is added for you.
/plugin install <plugin>@<marketplace> shows the plugin and asks for a scope: user (every project on your machine), project (shared through the repository) or local (you, in this repository).
/reload-plugins loads what you installed into the running session.
From a shell, claude plugin install <plugin>@<marketplace> --scope user does the same.
Rolling a plugin out to a team
Install scope is the team lever. A project-scope install is written to enabledPlugins in the repository's .claude/settings.json, so everyone who opens the project gets the same plugins. To publish your own, add a .claude-plugin/marketplace.json to a repository (a name, an owner and a list of plugins, each with a name and a source) and have the team add it once.
Review a plugin before it reaches the whole team: its hooks and MCP servers run as you, outside the sandbox, and every enabled plugin adds its skill and agent descriptions to Claude's context on every turn.
That context costs the most when nobody uses the plugin. In one engineer's 323 Claude Code sessions (September 2 to October 3, 2026), the Stripe plugin, enabled the whole time, used none of its 10 skills or its MCP tools.
A plugin you can read: Zest
Zest's own Claude Code plugin is a small example of the format: three hooks (SessionStart, SessionEnd and PostToolUse) that run a command from the plugin, and slash commands such as /zest:login and /zest:ignore. Install it with /plugin marketplace add https://github.com/Winding-Labs/zest-claude, then /plugin install zest, then /reload-plugins, then /zest:login.
From then on Zest records the skills and subagents each session uses, naming the ones that come from a plugin after it, next to the session's models, tokens and the pull request it led to. For a team, that shows which plugins earn their place in everyone's context and which nobody uses. More in Claude Code usage.
TRY IT ON YOUR TEAM
See what your team ships with AI.
Starter is free (3 seats, no credit card). Pro is $49 a month with 5 seats and Max is $149 a month with 10 seats — priced per workspace, not per engineer, with extra seats at $19 a month.
What is the difference between a plugin and a skill?+
A skill is one capability: a SKILL.md and its files. A plugin is the package: it can bundle many skills with agents, hooks and MCP servers, and installs, updates and disables as one unit.
Where are Claude Code plugins enabled?+
In the settings file for the scope you chose: user scope in ~/.claude/settings.json, project scope in the repository's .claude/settings.json, local scope in .claude/settings.local.json.
Are Claude Code plugins safe to install?+
Treat them like any code you run. A plugin's hooks and MCP servers run with your user permissions outside the sandbox, and auto-update (on by default for official marketplaces, off for third-party ones) can change files after you reviewed them.
