Guide

Claude Code plugins: how to install them, the official marketplace, and the most popular ones

One command installs a bundle of skills, hooks and tools. What's in Anthropic's marketplace, which plugins people actually use, and what to check before you install one.

A Claude Code plugin is a package of skills, slash commands, subagents, hooks, MCP servers or language servers that installs in one step. Type /plugin to browse: Claude Code adds Anthropic's official marketplace, claude-plugins-official, on first launch, with 315 plugins as of October 7, 2026. Install one with /plugin install name@claude-plugins-official, or add another marketplace with /plugin marketplace add owner/repo. The most installed are Anthropic's frontend-design (about 1.1 million installs), the community superpowers collection, and Anthropic's code-review. Anthropic warns that a plugin "can execute arbitrary code on your machine with your user privileges," so install only from sources you trust.

Key takeaways

  • What: a bundle of skills, commands, subagents, hooks, MCP and LSP servers, installed and updated as one unit.
  • Where: /plugin has Discover, Installed, Marketplaces and Errors tabs; the official marketplace also lists install counts on the web.
  • Popular: frontend-design, superpowers, code-review, context7, code-simplifier, playwright and github top the list.
  • Not free in context: every enabled plugin adds its skill and command descriptions to every turn; Claude Code shows a context cost for official plugins.
  • Teams: list a marketplace and plugins in the repository's .claude/settings.json, or lock them down with managed settings.
The most installed Claude Code plugins, from Anthropic's marketplace, October 7, 2026.
PluginMade byWhat it doesInstalls
frontend-designAnthropicProduction-grade interfaces1.13 million
superpowersJesse VincentDevelopment skills collection1.01 million
code-reviewAnthropicMulti-agent pull request review439,000
context7UpstashCurrent library docs418,000
code-simplifierAnthropicRefines code for clarity347,000
playwrightMicrosoftBrowser automation320,000
githubGitHubGitHub tools319,000
claude-md-managementAnthropicAudits CLAUDE.md287,000

What is a Claude Code plugin?

A folder Claude Code loads as one unit, usually with a manifest at .claude-plugin/plugin.json (Anthropic). It can contain skills, commands, subagents, hooks, MCP servers, language servers (LSP), output styles and more. A plugin's skills are namespaced, so a skill hello in my-plugin runs as /my-plugin:hello.

Plugins vs skills: a skill is one set of instructions (Claude Code skills). Anthropic suggests a plugin "when you want several skills, subagents, hooks, or MCP servers packaged as one unit," with one-command install and updates.

How do you install Claude Code plugins?

From Anthropic's guide (Anthropic):

  • Browse: type /plugin and open Discover. Official plugins show what they'll install and their context cost.
  • Install: /plugin install frontend-design@claude-plugins-official, then choose a scope: user (all your projects), project (.claude/settings.json, shared through git) or local.
  • Add a marketplace: /plugin marketplace add owner/repo (a GitHub repository, git URL or local path).
  • Manage: /plugin enable, disable and uninstall; /reload-plugins applies changes without restarting.
  • From the shell: claude plugin install, list, update and more.

Official marketplace plugins update automatically; community and third-party marketplaces don't unless you turn it on. Cloud sessions on claude.ai/code don't load your local plugins.

What are the best Claude Code plugins?

The ten most installed on Anthropic's marketplace, October 7, 2026 (Anthropic):

  1. frontend-design (Anthropic): distinctive, production-grade interfaces, about 1.13 million installs.
  2. superpowers (Jesse Vincent): a collection of development skills, about 1.01 million.
  3. code-review (Anthropic): multi-agent pull request review.
  4. context7 (Upstash): up-to-date library documentation.
  5. code-simplifier (Anthropic): refines code for clarity.
  6. playwright (Microsoft): browser automation.
  7. github (GitHub).
  8. claude-md-management (Anthropic): audits and updates CLAUDE.md.
  9. feature-dev (Anthropic): a feature workflow with exploration, architecture and review agents.
  10. security-guidance (Anthropic): flags risky code as Claude edits (Is Claude Code safe?).

Also popular: vercel, typescript-lsp, ralph-loop, commit-commands, figma, supabase and pr-review-toolkit. Note that "Anthropic verified" isn't the same as made by Anthropic: the badge also appears on vendors' plugins such as Slack and Adobe.

Which plugins reduce token usage?

The one Anthropic documents is code-intelligence (LSP) plugins: "A single 'go to definition' call replaces what might otherwise be a grep followed by reading multiple candidate files" (Anthropic). The official marketplace has them for 12 languages, including TypeScript, Python (pyright), Go, Rust, Java and C#; you also install the language server itself. No official plugin otherwise claims to save tokens, and plugins themselves cost context: check with claude plugin details <name> and disable ones you don't use. More in How to reduce Claude Code token usage.

How do teams share plugins?

Commit this to the repository's .claude/settings.json and teammates are offered the marketplace and plugin when they trust the folder (Anthropic):

{
  "extraKnownMarketplaces": {
    "team": { "source": { "source": "github", "repo": "your-org/plugins" } }
  },
  "enabledPlugins": { "tool@team": true }
}

Organizations can allow only listed marketplaces with strictKnownMarketplaces, block others with blockedMarketplaces, and push settings through managed policy.

How do you make a Claude Code plugin?

Create a folder with .claude-plugin/plugin.json (only name is required) and your skills, hooks or MCP config; check it with claude plugin validate, and test it with claude --plugin-dir ./my-plugin (Anthropic). To publish, add a .claude-plugin/marketplace.json to a git repository; anyone can then add it with /plugin marketplace add. Anthropic's plugin-dev plugin scaffolds one for you.

Are Claude Code plugins safe?

As safe as their author. A plugin "can execute arbitrary code on your machine with your user privileges"; its hooks and MCP servers run outside the sandbox, and with auto-update on, the files you reviewed can change (Anthropic). Before installing, read the details pane, the plugin's hooks/hooks.json and .mcp.json, and prefer Anthropic's and vendors' own plugins. Anthropic doesn't review third-party marketplaces.

Plugins in Codex and Cursor

  • Codex: one plugin directory serves ChatGPT and Codex; open it with /plugins in the Codex CLI (OpenAI).
  • Cursor: a reviewed Cursor Marketplace, plus team marketplaces on Teams and Enterprise (Cursor).

One setup for a whole team

A plugin listed in the repository still runs in each person's own session. In Poly (usepoly.co), a team works with one Claude Code or Codex agent in a shared browser room with one setup, everyone sees each step live, and any member can approve a change before it happens. Free to start. What is Poly?

Common questions

How do I install a Claude Code plugin?

Type /plugin and pick one from Discover, or run /plugin install name@claude-plugins-official and choose a scope. Add other marketplaces with /plugin marketplace add owner/repo.

What is the official Claude Code plugin marketplace?

claude-plugins-official, run by Anthropic and added automatically on first launch. It listed 315 plugins on October 7, 2026, from Anthropic and vendors such as GitHub, Microsoft, Vercel and Figma.

What is the difference between plugins and skills?

A skill is one set of instructions Claude loads when relevant. A plugin packages several skills, commands, subagents, hooks or MCP servers so they install and update together.

What are the most popular Claude Code plugins?

On Anthropic's marketplace in October 2026: frontend-design, superpowers, code-review, context7, code-simplifier, playwright, github, claude-md-management, feature-dev and security-guidance.

Are Claude Code plugins safe?

Only as safe as their source: Anthropic says a plugin can execute arbitrary code with your user privileges. Prefer Anthropic's and vendors' own plugins, and read a plugin's hooks and MCP config before installing.