The 5 Best Claude Code Marketplace Alternatives in 2026

Anthropic's official directory of Claude Code plugins, available as the claude-plugins-official repo and automatically present when you start Claude Code. Anthropic describes plugins as bundles that can include skills, agents, hooks, MCP servers, and other capabilities, so instead of rebuilding setups with custom commands, MCP configs, hooks, and project notes, teams can package workflows and install them as needed. You browse via the /plugin command's Discover tab or install directly with /plugin install github@claude-plugins-official, and refresh with /plugin marketplace update. The directory covers skills, agents, commands, hooks, and LSPs, and Anthropic positions it as a source of high-quality plugins. What it doesn't do is independently verify or benchmark the skills it lists, which matters when published research found roughly 26% of skills carry at least one vulnerability.

pricing See pricing

• + Zero setup: the official marketplace is automatically available when you start Claude Code
• + Anthropic-managed, so listings meet a quality bar
• + Covers the full plugin bundle: skills, agents, commands, hooks, and LSPs
• + One-command install and refresh workflow

• − No independent security scanning or vulnerability testing of listed skills
• − No benchmarking, so you can't tell which skills actually improve agent output
• − Smaller than community collections like the 34 collections indexed on aitmpl.com
• − Tied to Claude Code, with no equivalent loadout system for other agents

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 continuously crawls and dedupes skills from GitHub, then pins every version to a commit SHA, validates it, license-checks it, security-scans it, and quality-judges it. Every skill gets tested in a sandbox with the same model, with and without the skill, and SkillzBench publishes the resulting Lift in percentage points, always shown separately from ratings. A trust score from 0 to 100 combines security, quality, license, maintenance, and community signals, and money never touches scores. Per-agent loadouts let you toggle skills and tools per agent so each agent only sees what you chose. It also tracks upstream changes, re-scans, re-benchmarks, and notifies you. Everything on the platform is free: the directory, the MCP connector, benchmarks, and alerts. Skills work across Claude Code, Codex, Cursor, Gemini CLI, and chat-app agents.

pricing 0

• + Independent verification: security scans, license checks, and quality judging on every skill version
• + Actual benchmarks showing whether a skill helps, not just ratings
• + Works across multiple agents, not only Claude Code
• + Completely free, with ad revenue never touching scores or listings
• + Per-agent loadouts keep each agent's context clean

• − It's a directory and connector, not a plugin authoring platform, so you'll still install or author complex plugins elsewhere
• − Newer and less established than Anthropic's official directory
• − Benchmark coverage depends on what has been crawled and tested so far

https://skillzbench.com

aitmpl.com indexes the most popular Claude Code plugin collections, skill registries, and component marketplaces on GitHub, currently listing 34 collections. It's a discovery layer rather than a marketplace itself: you browse collections ranked by GitHub stars, including Ecc (229k stars), Claude Mem (87k), Claude Plugins Official (32k), Claude Hud (26k), and Claude Skills by alirezarezvani, which packages 345 skills, 30+ agents, and 70+ custom commands for Claude Code, Codex, Gemini CLI, Cursor, and 8 more coding agents. Each listing tags what the collection contains (skills, agents, commands, hooks, MCPs, LSPs) so you can filter by what you need. Because it aggregates rather than curates deeply, you get enormous variety, but no security scanning or benchmarking of the underlying skills.

pricing See pricing

• + Widest discovery surface: 34 collections in one browsable index
• + Star counts give a rough community-signal filter
• + Includes collections that work across Claude Code, Codex, Cursor, and other agents

• − Aggregator only: no security scanning, license checks, or benchmarks
• − Popularity is the sole ranking signal, and stars don't equal safety or quality
• − You still install each collection from GitHub yourself

https://www.aitmpl.com/plugins/

Composio's awesome-claude-plugins repo is a curated list of Claude Code plugins: slash commands, subagents, MCP servers, and hooks, installable as a third-party marketplace via /plugin marketplace add or a direct git clone. The collection spans code intelligence and LSP plugins (TypeScript, Pyright, Rust Analyzer, Go), MCP integration plugins for GitHub, Linear, Slack, Vercel, Sentry, and Supabase, code review plugins like PR Review Toolkit and CodeRabbit, and security plugins such as Security Guidance and Semgrep. Composio also ships a Connect MCP that lets Claude Code securely access apps like Jira, Linear, and GitHub from a single MCP. The curation is practical and well-organized by type, but it's a hand-picked list, with no independent scanning or benchmarks of the listed plugins.

pricing See pricing

• + Thoughtful curation by plugin type, with clear guidance on what to install first
• + Covers security plugins like Security Guidance and Semgrep, which many lists skip
• + Works as a third-party marketplace inside Claude Code's own plugin system

• − No independent security verification or benchmarking of listed plugins
• − The vendor's own Connect MCP is promoted alongside the open list, so the curation has a commercial angle
• − Maintaining many MCP servers at once can choke the LLM context window, as Composio itself warns

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

This open-source marketplace by jeremylongshore, listed on aitmpl.com, packages 425 plugins, 2,810 skills, and 200 agents for Claude Code, hosted at tonsofskills.com with its own ccpi CLI package manager. It's one of the largest single collections in the ecosystem, covering skills, agents, commands, and hooks. The dedicated CLI means you can install and manage packages outside Claude Code's built-in plugin system, which some teams prefer for scripting installs across many machines. The sheer volume is the selling point and also the risk: with nearly 3,000 skills and no described security scanning or benchmarking, you're trusting upstream authors, and published research finding roughly 26% of skills carry at least one vulnerability applies squarely to collections like this.

pricing See pricing

• + One of the largest skill and plugin collections available
• + Dedicated ccpi CLI for scripted, repeatable installs
• + Open source, so the whole marketplace is inspectable

• − No security scanning or benchmarking across its nearly 3,000 skills
• − Volume makes quality control hard, and bad skills are a documented ecosystem risk
• − A second package manager alongside Claude Code's own plugin system adds workflow complexity

https://tonsofskills.com

Superpowers Marketplace, by obra, is a curated Claude Code plugin marketplace listed on aitmpl.com with 1.1k stars, containing commands, MCPs, and skills. It installs through Claude Code's standard third-party marketplace mechanism, /plugin marketplace add, so it lives inside the same workflow as the official directory. Curation is the pitch: a smaller, hand-picked set rather than thousands of listings. That makes it easier to scan as a human, but there's no independent security scanning or benchmarking described, and its footprint is much smaller than the aggregators, so you'll likely pair it with another source for breadth.

pricing See pricing

• + Small and curated, so easy to evaluate by hand
• + Installs through Claude Code's native marketplace mechanism
• + Covers commands, MCPs, and skills together

• − Much smaller coverage than the big aggregators
• − No security scanning or benchmarking described
• − Single-maintainer curation, so updates depend on one person's attention

https://github.com/obra