The 5 Best Claude Code Marketplace Alternatives in 2026

Anthropic's official plugin directory for Claude Code. Plugins are packaged extensions that can bundle skills, agents, hooks, MCP servers, LSPs and other capabilities, replacing the old scatter of CLAUDE.md files, hook scripts and custom slash commands. The marketplace is automatically available when you start Claude Code: you run /plugin, browse the Discover tab, and install with one command such as /plugin install github@claude-plugins-official. The Anthropic-managed repo (claude-plugins-official) has around 32k GitHub stars, and Anthropic also publishes a separate Knowledge Work Plugins collection aimed at knowledge workers. Strengths are convenience and curation: everything installs in one place and Anthropic manages the directory. The limits are that it covers Claude Code only, and while Anthropic describes the collection as high quality, there's no published per-plugin security scan or benchmark result to check before you install.

pricing See pricing

• + Zero extra setup: the marketplace is available the moment you start Claude Code
• + Official Anthropic curation of the core directory
• + Plugins package skills, agents, hooks and MCP servers together, ending the era of scattered config files
• + Supports adding third-party marketplaces alongside the official one

• − Claude Code only; no coverage for Codex, Cursor or Gemini CLI
• − No published per-plugin security scans or benchmark results
• − Using it means committing to Claude Code, which has no free tier and pushes heavy agent users toward $100-200/month Max plans
• − Installing many MCP-heavy plugins can crowd the LLM context window

https://github.com/anthropics/claude-plugins-official

SkillzBench is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose. It exists because agent skills (SKILL.md instruction packages for Claude Code, Codex, Cursor, Gemini CLI and chat-app agents) have exploded in number while quality control hasn't: published research found roughly 26% of skills carry at least one vulnerability. The platform runs five pillars today. Discover continuously crawls and dedupes skills from GitHub. Verify pins every version to a commit SHA, then validates, license-checks, security-scans and quality-judges it. Benchmark tests every skill in a sandbox with the same model, with and without the skill, and publishes the Lift in percentage points. Equip gives you one free MCP server plus per-agent loadouts, so you toggle skills and tools per agent and the agent only sees what you chose. Stay current tracks upstream changes, re-scans, re-benchmarks and notifies you. Every skill carries a 0-100 trust score built from security, quality, license, maintenance and community signals, with Lift always shown separately. Everything is free, funded by blog advertising that never influences scores or listings.

pricing 0

• + Completely free, including the MCP connector and benchmarks
• + Only option here with published security scans and measured Lift per skill
• + Works across Claude Code, Codex, Cursor, Gemini CLI and chat-app agents
• + Per-agent loadouts prevent the context-window bloat that comes from loading every MCP tool at once
• + Independent: ad-funded, and ad money never affects scores or rankings

• − Not a plugin installer for Claude Code's native /plugin flow; it complements or replaces your skill sourcing rather than replacing Anthropic's curation
• − Younger and less well known than the official marketplace
• − Benchmark Lift figures vary by model and task, so results need interpreting per use case

https://skillzbench.com

Composio approaches the plugin problem from the integration side. Its Connect plugin installs a single MCP that lets Claude Code securely access apps like Jira, Linear, GitHub and more from one place, instead of wiring up a separate MCP server per tool. That matters because of a real problem the Composio write-up names plainly: install several MCP servers (Linear, GitHub, Supabase, Vercel) and you choke the LLM's context window as each loads all its available tools. Composio's answer is dynamic tool loading, where tools load as the context demands, plus a remote workbench for chaining multiple tools into automated workflows and a bash tool that lets the model write throwaway code to handle large files outside the context window. Setup is quick, reportedly about a minute via dashboard.composio.dev. It's the strongest pick when your bottleneck is connecting Claude Code to external services rather than finding skills.

pricing See pricing

• + Solves the multi-MCP context-window problem directly
• + One install covers many app integrations
• + Fast onboarding, about a minute to get started

• − Focused on integrations, not on vetting or benchmarking skills
• − No published per-skill trust scores or Lift data
• − Adds a third-party layer between your agent and your tools

https://composio.dev

Awesome Claude Plugins is the community-curated, open-source collection maintained under Composio's GitHub organization. It gathers slash commands, subagents, MCP servers and hooks for Claude Code into one browsable repo, and it doubles as a third-party marketplace you can add directly inside Claude Code with /plugin marketplace add owner/repo-name. You can also clone the repository and install from it as a local plugin directory. It's a natural first stop beyond the official marketplace because it's free, transparent (everything is in the open on GitHub), and curated by active Claude Code users rather than a vendor's marketing team. What it doesn't do is verify: there are no published security scans, license checks or benchmark results for the items it lists, so you're trusting community curation and your own review.

pricing See pricing

• + Free and fully open source
• + Community-curated, so popular items get real-world filtering
• + Installs natively inside Claude Code as a marketplace

• − No security scanning or quality benchmarks on listed items
• − Claude Code focused
• − Curation quality varies item by item

https://github.com/composiohq/awesome-claude-plugins

This collection is the volume play: 345 Claude Code skills and agent skills, including 30+ agents, 70+ custom commands and 330+ skills with customizable references and scripts. Its distinguishing trait is breadth of audience and platform: it targets Claude Code, Codex, Gemini CLI, Cursor and eight more coding agents, and the skills span engineering, marketing, product, compliance, C-level advisory, research, business operations, commercial and finance, plus daily productivity. For a small business owner who wants an agent to help with marketing or finance work rather than just code, that coverage is unusual and useful. The trade-off is trust: with roughly 26% of agent skills in published research carrying at least one vulnerability, a 345-item unvetted collection is exactly the kind of source where you'd want independent security scanning and benchmarking before switching anything on.

pricing See pricing

• + Unusual breadth beyond coding: marketing, compliance, finance and operations skills
• + Cross-agent compatibility with 12+ coding agents
• + Free and open on GitHub

• − No published vetting, security scans or benchmarks
• − Sheer volume makes quality uneven and discovery slow
• − You're trusting one maintainer's curation

https://github.com/alirezarezvani/claude-skills

This open-source marketplace goes even bigger on volume: 425 plugins, 2,810 skills and 200 agents for Claude Code, hosted at tonsofskills.com with its own ccpi CLI package manager for installing from the terminal. If the official marketplace feels thin for your use case, this is the kind of catalog that guarantees you'll find something adjacent to what you need. The ccpi CLI is a genuine convenience for teams that prefer managing plugins outside the /plugin flow. The same caveat applies with more force at this scale: 2,810 skills is far beyond what any team can review by hand, there's no published security or quality data per item, and with about a quarter of agent skills in published research carrying at least one vulnerability, bulk installs are where the real risk lives.

pricing See pricing

• + Largest raw catalog in this comparison
• + Own CLI package manager simplifies bulk management
• + Free and open source

• − No published vetting at any level
• − Volume without curation makes discovery a chore
• − Claude Code only

https://tonsofskills.com

Buildwithclaude is a single hub for finding Claude skills, agents, commands, hooks, plugins and marketplace collections, covering Claude Code, Claude Desktop, the Agent SDK and OpenClaw. Think of it as a map of the ecosystem rather than a catalog itself: it indexes what's out there, including other marketplace collections, so it's genuinely useful for discovery when you don't know what exists. Around 3.2k GitHub stars reflects a modest but real audience. Because it aggregates rather than produces, it inherits the weaknesses of whatever it links to, and it adds no verification layer of its own. It pairs well with a vetting tool: use Buildwithclaude to find candidates, then check them against security scans and benchmarks before installing.

pricing See pricing

• + Good discovery starting point across the Claude ecosystem
• + Free and open
• + Broader product coverage than most collections, including Claude Desktop and the Agent SDK

• − Aggregates rather than verifies; no security or quality layer
• − Smaller community than the official marketplace
• − Quality depends entirely on the linked sources

https://github.com/davepoon/buildwithclaude