Browse the full directory of Claude Code plugins — commands, agents, skills, MCP servers, and more.
Browse plugins →32,019 active plugins, 282,325 components, 13,151 authors — a data-driven breakdown of the Claude plugin ecosystem in April 2026, from component distribution to install trends and category landscape.
The Claude plugin ecosystem has reached 32,019 active plugins built by 13,151 authors, collectively offering 282,325 components. These aren't vanity numbers — 18,472 installs occurred in the last 30 days, and 2,069 new plugins were published just this week, roughly matching last week's 2,084 additions.
That sustained pace of over two thousand new plugins per week signals an ecosystem that has moved past the initial land-grab phase into steady, sustained growth. The week-over-week delta is essentially flat, which at this scale is a sign of maturity rather than stagnation.
Not all components are created equal. Here's how the 282,325 components break down across the nine supported types:
| Component Type | Count | Share |
|---|---|---|
| Skills | 161,540 | 57.2% |
| Commands | 58,121 | 20.6% |
| Agents | 48,012 | 17.0% |
| MCP Servers | 7,159 | 2.5% |
| Hooks | 6,723 | 2.4% |
| LSP Servers | 627 | 0.2% |
| Output Styles | 92 | — |
| Themes | 42 | — |
| Monitors | 9 | — |
Skills account for more than half of all components, which reflects their nature as the most granular unit of capability in Claude Code. A single plugin can bundle dozens of skills covering different aspects of a workflow. everything-claude-code, for example, ships 183 skills alongside 48 agents and 79 commands — a single plugin contributing hundreds of components.
Commands and agents together make up another 37.6% of the ecosystem. Commands provide direct user-invokable actions (think slash commands), while agents orchestrate multi-step workflows autonomously. The relatively even split between them suggests developers value both explicit control and autonomous operation in their Claude Code workflows.
The long tail reveals emerging integration patterns. MCP servers (7,159) and hooks (6,723) show that developers are connecting Claude Code to external tools and automating lifecycle events — but at just 2.5% and 2.4% of components respectively, these integration points remain underexplored relative to the core skill/command/agent triad. LSP servers (627), output styles (92), themes (42), and monitors (9) are specialized types still finding their audience.
Install velocity — the number of installs in the past 7 days — tells a different story than star counts. Here are the plugins gaining traction right now.
superpowers leads with 1,730 installs in the last 7 days and 169,218 stars. It enforces strict TDD workflows, generates multi-step implementation plans, and executes them in isolated git worktrees with batch reviews and verifications. It bundles commands, agents, skills, and hooks — dispatching parallel agents for independent tasks and verifying completeness via tests, linters, and builds.
caveman follows with 787 installs over 7 days and 46,962 stars. Its value proposition is concrete: 75% token savings by switching Claude to a compressed communication mode. It bundles skills and hooks that auto-generate compressed Conventional Commits, terse PR reviews, and condensed CLAUDE.md files while maintaining technical accuracy across lite, full, and ultra compression levels.
everything-claude-code (295 installs over 7 days, 169,421 stars) takes a maximalist approach. Its 48 specialist agents, 183 skills, and 79 commands span 20+ tech stacks, with hooks for quality enforcement and autonomous loops for continuous development. It covers commands, agents, skills, hooks, and MCP — all five core component types.
The next tier of install leaders reveals diverse use cases:
Star counts paint a partially different picture. The two highest-starred plugins — everything-claude-code (169,421) and superpowers (169,218) — also lead installs. But prompts.chat (159,878 stars), Vercel's cache-components (139,217 stars), and Vercel's next-js plugin (139,179 stars) carry massive star counts from their parent repositories without necessarily matching that level of direct install velocity. Stars reflect community awareness; installs reflect active adoption.
The plugin ecosystem segments into clear categories, each reflecting a distinct developer need:
Utilities commanding 38.7% of all plugins reflects a natural pattern: developers first optimize their existing Claude Code workflows before adding entirely new capabilities. The productivity and deployment categories running neck-and-neck at 7,375 and 7,224 plugins respectively suggests balanced investment in both planning-and-execution tools and ship-and-deploy tooling.
Three patterns emerge from the data that matter if you're building or choosing plugins:
Multi-component plugins correlate with higher adoption. The install leaders tend to bundle multiple component types. superpowers ships commands, agents, skills, and hooks. everything-claude-code covers all five core types plus MCP. Single-type plugins can succeed — andrej-karpathy-skills proves that with 102 installs over 7 days on skills alone — but the install velocity leaders are generally comprehensive toolkits rather than narrow utilities.
Hooks and MCP are the integration frontier. With only 6,723 hooks and 7,159 MCP servers in an ecosystem of 282,325 components, these integration points sit at 2.4% and 2.5% respectively. Plugins that connect Claude Code to external systems — whether through lifecycle hooks or MCP tool servers — have room to grow without competing in the crowded skills space where 161,540 components already exist.
Concrete outcomes drive installs. The two highest-velocity plugins solve specific, measurable problems: superpowers enforces TDD discipline with verified completeness, caveman reduces token costs by 75%. Neither promises vague AI augmentation — they deliver defined outcomes. Plugin developers looking for adoption should target a clear pain point with quantifiable results rather than broad capability.
With 13,151 authors contributing and 2,069 new plugins arriving each week, the ecosystem has critical mass. For new entrants, the differentiator is increasingly integration depth and measurable impact rather than capability breadth.
Browse the full ecosystem at the plugins directory, or explore specific component types: commands, agents, skills, and MCP servers.
Enforce TDD workflows with red-green-refactor cycles, generate and execute batched implementation plans via subagents, dispatch parallel agents for independent task fixes like test failures, perform technical code reviews with git diffs, debug root causes before fixes, create isolated git worktrees for features, and verify tests/builds before merging or PRs in Claude Code sessions.
Adopt caveman-style terse communication in Claude Code sessions to cut token usage by 75% while retaining full technical accuracy, enabling workflows like delegating to compressed subagents for code location and 1-2 file edits, generating conventional git commits, delivering actionable diff reviews, compressing docs, and tracking real token stats.
Supercharge Claude Code with 300+ agents, skills, commands, and hooks to orchestrate autonomous multi-agent coding workflows, enforce TDD, conduct security audits, generate production code across JS/TS/Python/Rust/mobile stacks, optimize performance, and automate deployments/testing.
Access 50+ UI/UX styles, color palettes, font pairings, guidelines, and charts to plan, build, review, and optimize web and mobile interfaces across React, Next.js, Vue, Svelte, Tailwind, React Native, and Flutter.
Design, audit, optimize, and iterate production-grade frontend interfaces like websites, dashboards, components, and forms using 23 AI commands for UX review, accessibility checks, performance tuning, responsive layouts, theming, typography, motion, and anti-pattern detection.
Automate browser tasks via Bash or CLI commands: navigate pages, interact with elements via clicks/fills/types, capture screenshots, extract text/HTML for web testing, form submission, data scraping, Electron app control, Slack workflows. Deploy on Vercel or AWS Bedrock for AI agents.
Apply Andrej Karpathy-inspired rules to guide code writing, reviews, and refactoring: enforce simplicity, surgical changes, and verifiable success criteria to sidestep common LLM coding pitfalls like overcomplication.
Monitor Claude Code sessions in real-time via a statusline HUD displaying context health, tool activity, agent status, and todo progress. Interactively configure HUD layout, presets, display elements like git status and usage, saving to config.json. Detect and clean ghost plugin installations by scanning cache, registry, and temp files on macOS, Linux, or Windows.
Search, retrieve, improve, and manage thousands of AI prompts and Claude skills from prompts.chat directly in your coding assistant. Install skills to extend capabilities, fill prompt variables, save custom prompts with metadata, and enhance them using AI.
Implement Next.js Cache Components and Partial Prerendering (PPR) in projects with cacheComponents enabled, applying 'use cache' directives, cacheLife(), cacheTag(), revalidateTag() for invalidation, static/dynamic optimization, and cache debugging patterns.
Detect and auto-fix Next.js compilation errors after code edits by running a Turbopack dev server. Replaces manual `next build` commands, enabling faster debugging of runtime issues, performance problems, and edge cases in dev and production.
Add persistent memory to Claude Code tasks and AI apps via Mem0: retrieve relevant past decisions, strategies, and session states on new tasks; store user data for personalization; enable semantic search across long-term memories using Python/TS SDKs, hooks, and MCP tools.
Accelerate full-stack development by invoking specialized AI skills to generate, debug, optimize code across dozens of languages and frameworks; design scalable architectures, DevOps pipelines, secure systems; perform code reviews, testing; and automate Jira workflows for epic planning, ticket execution, sprints, retrospectives, and Confluence docs.
Automate browsers and run end-to-end tests with Playwright directly in Claude. Interact with web pages by clicking elements, filling forms, taking screenshots, generating traces, and executing testing workflows locally via npx subprocess.