Best of · updated October 2026

Best Claude Code Plugins in 2026

Twelve plugins that bundle skills, subagents, hooks or MCP servers for Claude Code, and what each one adds.

Edited by hand at Open Skills

How we picked

We have not yet tested these plugins first-hand, and this list does not claim otherwise. Each one was chosen on things we can check: how widely it is used and searched for, whether its repository is still being maintained, and what it adds according to its own manifest and documentation. We confirmed that each repository and plugin folder linked here exists.

A plugin is not the same as a skill. A Claude Code plugin is a directory that can contain skills, subagents, hooks and MCP servers, installed as one unit. Ten of these are listed in Anthropic's official marketplace, claude-plugins-official, which Claude Code adds automatically. The last two, claude-mem and ECC, are installed from their authors' own marketplaces.

Two cautions from Anthropic's documentation apply to all of them. An enabled plugin is part of every session, so its skill and agent descriptions use context even when you do not call them. And a plugin can run hooks and MCP servers with your user privileges, so read what it adds before you install it.

For the difference between plugins, skills and subagents, read Claude skills vs agents, plugins and commands.

Superpowers

Jesse VincentDevelopment294k

A full software development workflow for coding agents, built from composable skills for brainstorming, planning, TDD, debugging and review.

Best for: a full development method in one install.

$ npx skills add obra/superpowers

feature-dev

Anthropic

A feature development workflow with specialised agents for exploring the codebase, designing the architecture and reviewing quality.

Best for: building a feature in an unfamiliar codebase, step by step.

View feature-dev on GitHub

code-review

Anthropic

Automated code review for pull requests. It uses several specialised agents and scores findings by confidence to filter out false positives.

Best for: a second pass over a pull request before a person reads it.

View code-review on GitHub

pr-review-toolkit

Anthropic

A group of review agents, each focused on one area, including comments, tests, error handling, type design, code quality and simplification.

Best for: reviews where you want one aspect checked at a time.

View pr-review-toolkit on GitHub

commit-commands

Anthropic

Commands for git workflows, covering commit, push and pull request creation.

Best for: people who commit and open pull requests from inside Claude Code.

View commit-commands on GitHub

skill-creator

Anthropic

Creates new skills, improves existing ones and measures skill performance with evals and benchmarks.

Best for: writing your own skills and testing them.

View skill-creator on GitHub

hookify

Anthropic

Creates custom hooks that prevent unwanted behaviour, either by analysing a conversation or from explicit instructions. Rules are defined in Markdown files.

Best for: turning a correction you keep repeating into a rule that always applies.

View hookify on GitHub

security-guidance

Anthropic

Security review for code Claude writes, with pattern-based warnings on edits, a diff review when Claude stops, and a commit reviewer that looks for injection, XSS, SSRF and hardcoded secrets.

Best for: catching common vulnerabilities as the code is written.

View security-guidance on GitHub

plugin-dev

Anthropic

A toolkit for developing Claude Code plugins, with skills covering hooks, MCP integration, commands and agents.

Best for: building and validating a plugin of your own.

View plugin-dev on GitHub

context7

Upstash

Connects Claude Code to Context7's hosted MCP server, which pulls version-specific documentation and code examples for libraries into context.

Best for: working with libraries whose APIs have changed recently.

View context7 on GitHub

claude-mem

thedotmack

Persistent context across sessions. It captures what the agent does during a session, compresses it, and injects relevant context into later sessions.

Best for: long projects where the agent should remember earlier sessions.

View claude-mem on GitHub

ECC (Everything Claude Code)

affaan-m

A large collection of skills, memory, security and research-first development practices for Claude Code, also packaged for Codex, OpenCode and Cursor.

Best for: people who want a broad, opinionated setup to start from and trim down.

View ECC (Everything Claude Code) on GitHub

Install them

$ /plugin install <plugin>@claude-plugins-official

Run it inside Claude Code. A plugin from another marketplace needs /plugin marketplace add <owner>/<repo> first; its README has the exact commands.