14 Best Claude Code Plugins in 2026

How to install Claude Code plugins with /plugin, plus the 14 best for 2026: features, drawbacks, and who each one fits, checked against Anthropic docs.

Book a Free Strategy Call

Skip the read: talk to the ManasFlow team in 30 min.

Free strategy call. We map your AI engineering team, you keep the notes.

The best Claude Code plugins in 2026 are Context7 for live version-specific docs, Frontend Design for UI that looks shipped, Superpowers for structured TDD and sub-agent workflows, Chrome DevTools MCP for real-browser debugging, and Anthropic's LSP pack for accurate code intelligence. Round out the stack with the GitHub, Playwright, Semgrep, Linear, and Vercel plugins and you cover repo workflows, E2E tests, security scanning, backlog sync, and deploys.

Last checked: October 2026 (every plugin re-checked against Anthropic's marketplace listings and the Claude Code docs on October 21, 2026).

Claude Code changed the way teams ship software in 2025. By 2026, the question is no longer whether to use it. It is which plugins to bolt on so your agent stops guessing and starts shipping. Anthropic's official marketplace now lists more than 300 plugins (315 entries in its marketplace.json on GitHub when we checked on October 21, 2026), most of them from partners and tool vendors rather than Anthropic itself.

The hard part is separating the plugins that meaningfully change the loop from the ones that just add commands to your terminal. Most plugins do not survive a week of real use. A few become non-negotiable because they reduce hallucinations, anchor Claude to live data, or replace whole categories of manual review. This is the difference between a curiosity install and a plugin that earns a permanent slot in your team's ~/.claude config.

This guide compares the 14 best Claude Code plugins in 2026. Real features, honest pricing where it is publicly known, pros and cons, and a framework to pick the right stack for your codebase.

What are Claude Code plugins and how do you install one?

Claude Code plugins are packages that add skills, agents, hooks, and MCP servers to Claude Code as one unit. To install one, run /plugin in a session and pick it from the Discover tab, or type /plugin install name@claude-plugins-official. Then choose a scope: user, project, or local.

Claude Code adds the official marketplace, claude-plugins-official, the first time you start an interactive terminal session. For a plugin from any other marketplace, add the marketplace first:

From a shell or a setup script, claude plugin install plugin-name@marketplace-name does the same job, and claude plugin list shows what is installed. Anthropic runs three general marketplaces: official (claude-plugins-official), community (claude-community, added with /plugin marketplace add anthropics/claude-plugins-community), and a demo set (claude-code-plugins). Anything else is third-party.

A plugin can run hooks and MCP servers with your user privileges, so read its details pane before you pick a scope. Our guide to Claude Code security risks covers what to check, and if you are starting from zero, how to set up Claude Code comes first. Install steps checked against the Claude Code docs on October 21, 2026.

Best Claude Code plugins: a brief overview

• Context7: Best for up-to-date library documentation injected into prompts.
• Frontend Design: Best for shipping UI that does not look like a 2023 Bootstrap template.
• Superpowers: Best for structured workflows: TDD, brainstorming, sub-agent driven dev.
• Chrome DevTools MCP: Best for live frontend debugging through the real browser.
• Anthropic Language Servers (LSP pack): Best for accurate code navigation and diagnostics across 12 language servers.
• GitHub Plugin: Best for PR review, issue triage, and repo context inside Claude.
• Playwright Plugin: Best for end-to-end test authoring and execution by the agent.
• Semgrep Plugin: Best for explicit static analysis and security scanning steps.
• Linear Plugin: Best for connecting Claude Code to your actual product backlog.
• Vercel Plugin: Best for deploy-aware agents that read logs and ship to prod.
• HOL Plugins: Best for cross-platform plugin discovery with built-in security scanning.
• HOL Guard: Best for pausing risky agent actions before they execute.
• Code Review: Best for a multi-agent first-pass review of pull requests.
• Ralph Loop: Best for letting Claude iterate on a well-specified task until it is done.

Free weekly brief

Steal our production automations

The exact n8n flows, Claude Code setups, and prompts we ship for clients, broken down step by step. No spam, unsubscribe anytime.

1. Context7, best for live documentation

Context7 is the plugin most experienced Claude Code users install first. It is an MCP-backed plugin that pulls version-specific documentation straight from source repositories and injects it into the model's context window. Instead of Claude guessing the current React 19 or Next.js 16 API based on training data, it reads the actual docs at the moment you ask. It is built by Upstash and connects to Context7's hosted remote MCP server.

For teams shipping against fast-moving libraries (Next.js, Tailwind, LangChain, the Vercel AI SDK, Supabase), Context7 cuts an entire class of hallucinations. The tradeoff is latency: every doc lookup adds tokens and a network round-trip.

Key features

• Pulls live documentation from upstream repos and registries
• Version-aware: you can pin which release to read against
• MCP-based, works in Claude Code, Claude Desktop, and Cowork
• Library catalog covers most major JS/TS, Python, Go, and Rust frameworks
• Zero-config: install and reference libraries by name in your prompt

Best for

• Teams working with libraries that change quickly
• Anyone tired of fixing fabricated method signatures
• Projects pinned to specific framework versions

Pricing

• Free

Pros

• Eliminates the most common category of Claude hallucination
• Drop-in, no config files to maintain
• Works alongside any other MCP server

Cons

• Adds token cost to every doc lookup
• Coverage is broad but not exhaustive; niche libraries may not be indexed

2. Frontend Design, best for UI output that looks shipped

Frontend Design is Anthropic's own plugin, listed as Anthropic verified on Claude Marketplace. It exists because raw Claude Code produces functional but generic UI. Frontend Design wires the agent into design tokens, screenshots, layout reasoning, and a curated set of UI patterns so the components it ships actually look intentional.

For agencies and product teams that care about visual polish, this plugin is the difference between Claude generating a wireframe and Claude generating something a designer would not immediately rewrite. It pairs naturally with Chrome DevTools MCP for verifying the rendered output.

Key features

• Design-token aware: respects your existing Tailwind config and CSS variables
• Screenshot-first workflow: Claude sees what it ships
• Built-in component patterns for common UI surfaces
• Works with Shadcn, Radix, and most modern React component libraries
• First-party, maintained by Anthropic

Best for

• Frontend teams that care about visual quality
• Agencies shipping client-facing products
• Solo founders without a designer

Pricing

• Free

Pros

• Anthropic verified on Claude Marketplace
• Backed by Anthropic, not a hobby project
• Composes well with Chrome DevTools and Playwright

Cons

• Adds context overhead even when you only want logic changes
• Cannot replace an actual designer for novel visual systems

3. Superpowers, best for structured workflows

Superpowers reframes Claude Code from a single agent to a small dev team. It ships a curated set of workflows (brainstorming, sub-agent driven development, systematic debugging, red/green TDD, code review) plus tooling to author and test your own skills. It comes from an independent author (obra, github.com/obra/superpowers) and is listed in the official marketplace.

The value of Superpowers is structural. It nudges the agent into patterns that produce better code: write the test first, run the test, fix the failure, review the diff. If your team has tried Claude Code and felt like the agent rushes to a solution, Superpowers is the corrective. For the pick of that stack, see the 10 best Claude Code sub-agents.

Key features

• Sub-agent orchestration with automatic code review handoff
• Built-in TDD loop with red/green/refactor enforcement
• Brainstorming and planning workflows for non-trivial features
• Skill authoring framework: write, test, and ship your own
• Systematic debugging recipes

Best for

• Teams that want consistent agent behavior across engineers
• Anyone moving from "ask Claude to fix a bug" to multi-step delivery
• Skill authors building reusable workflows

Pricing

• Free, open source

Pros

• Battle-tested by a large community
• Forces good habits: test first, review before merge
• Skill framework is genuinely useful for internal tooling

Cons

• Opinionated; fights you if you do not want a TDD loop
• Higher token usage than ad-hoc Claude Code use

4. Chrome DevTools MCP, best for browser-level debugging

Chrome DevTools MCP (chrome-devtools-mcp in the official marketplace, published by the Chrome DevTools team) gives Claude direct access to a running browser via the Chrome DevTools Protocol. The agent can read network requests, inspect console errors, dump the DOM, run scripts in the page context, and screenshot what the user actually sees. For frontend bugs that only reproduce in the browser, this plugin is the difference between Claude guessing and Claude debugging.

It pairs with Frontend Design to close the loop: Frontend Design helps ship UI, Chrome DevTools verifies it works. This is the recommended stack for any team shipping web frontends with Claude Code.

Key features

• Live network inspection (requests, responses, timing)
• Console log and error capture
• DOM querying and manipulation
• In-page JavaScript execution
• Screenshot capture at any viewport size

Best for

• Frontend developers debugging real browser issues
• Anyone working on web performance
• Teams that pair Claude with manual QA

Pricing

• Free

Pros

• Eliminates the "works on my machine" guesswork
• Pairs perfectly with Frontend Design and Playwright
• Zero dependencies beyond a Chromium browser

Cons

• Requires a local browser instance, heavier than pure MCP plugins
• Not useful for non-web codebases

5. Anthropic Language Servers (LSP pack), best for multi-language code intelligence

The official LSP pack bundles 12 language servers into Claude Code: TypeScript, Python, Go, Rust, C/C++, Java, C#, Kotlin, PHP, Ruby, Lua, and Swift. With LSPs active, Claude gets the same code intelligence your IDE has: real types, real symbol resolution, real diagnostics. This is the foundation plugin: install it before anything else if you work in a typed language.

Each language is a separate plugin in the official marketplace (for example typescript-lsp, pyright-lsp, gopls-lsp, rust-analyzer-lsp), so install only the ones for your stack.

Key features

• 12 language servers maintained by Anthropic
• Automatic diagnostics surfaced into Claude's context
• Symbol navigation: "go to definition," find references, rename
• Type-aware refactoring
• Works in any Claude Code project, no config

Best for

• Teams on typed languages (TS, Go, Rust, Java, etc.)
• Large codebases where symbol search matters
• Anyone doing refactors

Pricing

• Free, first-party

Pros

• Single biggest accuracy upgrade for typed-language work
• Maintained by Anthropic, not a community fork
• Zero config

Cons

• Minimal value for untyped or dynamic-only codebases
• LSPs can be memory-heavy on big monorepos

6. GitHub Plugin, best for repo and PR workflows

The GitHub plugin connects Claude Code to your repositories, pull requests, issues, and Actions. The agent can read PR diffs, leave reviews, triage issues, search across repos, and run workflows. For any team using GitHub as the source of truth, which is most teams, this plugin removes the need to copy-paste between terminal and browser.

It is the most common second-install after the LSP pack, and Anthropic actively maintains it as a partner plugin.

Key features

• PR diff reading and review comment posting
• Issue search, creation, and triage
• Cross-repo code search
• GitHub Actions trigger and log inspection
• Branch and commit operations

Best for

• Engineering teams doing structured code review
• Open-source maintainers triaging issues
• Anyone who lives in PRs

Pricing

• Free plugin (your existing GitHub plan applies)

Pros

• Tight integration with the GitHub API surface
• Lets Claude review PRs as a first pass before humans
• Maintained by GitHub as an Anthropic partner

Cons

• Permissions setup requires care; wide tokens are a liability
• Not a replacement for human PR reviewers on critical changes

7. Playwright Plugin, best for end-to-end test authoring

The Playwright plugin lets Claude Code author, run, and debug end-to-end tests against your app. Combined with Chrome DevTools MCP and Frontend Design, it closes the loop on UI changes: Claude builds, screenshots, verifies, then writes a regression test before the diff is committed.

Playwright is also one of the few plugins where the ROI is easy to measure: test coverage written by the agent that would have been deferred or skipped manually.

Key features

• Generates Playwright tests from natural language specs
• Cross-browser support (Chromium, Firefox, WebKit)
• Trace and video capture on failure
• Run tests headless or headed
• Integrates with most CI providers out of the box

Best for

• Teams without dedicated QA engineers
• Apps that ship without enough E2E coverage
• Frontend agencies handing off projects

Pricing

• Free