Awesome Claude Code: 11 Curated Lists Worth Bookmarking

An awesome-list for Claude Code is a curated, community-maintained directory of skills, hooks, slash commands, agents, MCP servers, and plugins shipped outside Anthropic's official docs. Eleven lists lead the 2026 ecosystem covering different scopes: aggregators (141k stars), curated indexes (36.8k stars), MCP-specific catalogs, design references, and adversarial reverse-engineering of competing AI tools. Awesome-lists are discovery infrastructure; ClaudeFast Kits are the assembled production output.

Quick verdict: hesreallyhim/awesome-claude-code (36.8k stars) is the canonical hand-curated list. punkpeye/awesome-mcp-servers covers MCP. affaan-m/everything-claude-code (141k) is the aggregator firehose. Most teams overspend on browsing lists and underspend on shipping. ClaudeFast Code Kit ships the assembled production stack so you can stop curating.

Quick Win: Bookmark These Three First

In order of priority for most engineers:

1. hesreallyhim/awesome-claude-code for skills, hooks, slash commands, agents
2. punkpeye/awesome-mcp-servers for MCP servers across the entire ecosystem
3. affaan-m/everything-claude-code for the unfiltered firehose

Below, the eight other lists that are worth your bookmark slot, plus three you can safely skip.

Why Awesome-Lists Multiplied in 2026

Three forces reshaped the list ecosystem in 2026. quemsah/awesome-claude-plugins indexed 15,134 Claude Code plugin repositories by May 1, 2026, up from roughly 4,000 a year earlier. obra/superpowers crossed 94,000 stars and was officially accepted into the Anthropic skills marketplace, signalling that community frameworks are now first-class citizens. And Anthropic's April 4, 2026 policy change blocked Pro and Max subscribers from third-party agent frameworks, sharpening the line between official paths and community add-ons.

The result: more lists than any single engineer can read. Three filters separate signal from noise:

1. Recent commits: A list updated yesterday is current. A list updated a year ago is archeology.
2. Practitioner curation: Lists maintained by an SEO, a marketer, an engineer who actually ships are sharper than lists maintained by list-maintainers.
3. Scoped, not exhaustive: A list of 50 tools the curator personally tested beats a list of 5,000 tools auto-scraped from GitHub.

1. affaan-m/everything-claude-code (141k stars)

The largest aggregator by raw count. Comprehensive collection of Claude Code patterns, resources, tips, and configurations from the community.

Why it's #1 by stars: It's effectively the un-curated index. Useful when you want to see everything that exists, less useful when you need to pick one tool.

Best for: Discovery passes. When you're not sure what category you need, start here.

2. hesreallyhim/awesome-claude-code (36.8k stars)

Curated list of skills, hooks, slash-commands, agent orchestrators, applications, and plugins for Claude Code. Maintained Python repository.

Why it's the canonical reference: hesreallyhim curates with judgment. Tools that don't work get cut. Tools that do work get tagged with what they're actually for.

Best for: Picking one tool per category. If you only bookmark one Claude Code awesome-list, this is it.

3. quemsah/awesome-claude-plugins (recent)

Automated collection of Claude Code plugin adoption metrics across GitHub repositories using n8n workflows. Last updated May 1, 2026 with 15,134 indexed repositories.

Why it's interesting: It's auto-discovered, not human-curated. That means it catches new repos within hours of publication, but you have to filter the signal yourself.

Best for: Catching new tools before they hit the human-curated lists. Use it as the early-warning feed.

4. wong2/awesome-mcp-servers

Comprehensive curated collection of Model Context Protocol servers. Spans file systems, databases, APIs, cloud platforms, and developer tools.

Why it's separate from awesome-claude-code: MCP servers are protocol-agnostic and used by Claude Code, Claude Desktop, Cursor, and others. Keeping the MCP list separate keeps both lists scannable.

Best for: When you need to integrate Claude with a specific external system. Search the list for the platform name (Stripe, Notion, Linear).

5. punkpeye/awesome-mcp-servers

Second large MCP servers list. Overlaps with wong2's list but with different curator judgment.

Why bookmark both: Different curators catch different repos. Cross-reference both when researching a specific MCP integration.

Best for: Dual lookup. Check both before assuming an MCP doesn't exist.

6. ComposioHQ/awesome-claude-skills

Curated skills list with progressive-disclosure architecture explanations. Includes pre-built workflow skills for 78 SaaS apps via Rube MCP.

Why it's here despite the smaller star count: The 78-SaaS workflow library is unique. Most skills repos are bring-your-own-integration. This one ships with integrations to common SaaS already wired.

Best for: Anyone whose stack includes mainstream SaaS (Slack, Notion, Linear, GitHub). Skip if you're on niche or self-hosted tools.

7. VoltAgent/awesome-design-md (18.3k stars)

Collection of 55+ DESIGN.md files extracted from popular websites (Claude, Stripe, Vercel, Linear) that AI coding agents use to replicate production-quality UI design systems.

Why it's unusual: Not a tools list, a reference list. Each DESIGN.md is a working design system spec that you can hand to Claude Code as context. The Stripe and Linear specs alone are worth more than most paid design system documentation.

Best for: Frontend work where you want production-grade UI without reverse-engineering every component yourself.

8. VoltAgent/awesome-agent-skills (14.4k stars)

1,000+ agent skills from official dev teams and community, compatible with Codex, Antigravity, Gemini CLI, Cursor, and Claude Code.

Why it's broader than awesome-claude-code: Cross-platform. If you also use Cursor or Gemini CLI, this list shows you skills that work across all three so you don't fork your workflow per tool.

Best for: Multi-tool engineers. If you only use Claude Code, hesreallyhim's list is tighter.

9. Shubhamsaboo/awesome-llm-apps (104.6k stars)

Collection of LLM apps with AI agents and RAG using OpenAI, Anthropic, Gemini, and open-source models. Python.

Why it's here: When you need to see how a real app uses LLMs end to end, not just one skill in isolation. The RAG and agent patterns translate directly to Claude Code design decisions.

Best for: Reference architectures. Read it before designing your own agent or RAG system, not before installing your next tool.

10. x1xhlol/system-prompts-and-models-of-ai-tools (134.5k stars)

System prompts and internal tool definitions from Augment Code, Claude Code, Cursor, Devin AI, Junie, Kiro, Lovable, Manus, NotionAI, Perplexity, Replit, VSCode Agent, Warp.dev, Windsurf, Xcode, Z.ai Code, and more.

Why it's here despite being adversarial: Knowing how the leading AI coding tools actually prompt their models is the highest-signal calibration data on the planet for anyone writing skills, agents, or CLAUDE.md files.

Best for: Reading once. The patterns will improve every prompt you write afterward.

11. shanraisshan/claude-code-best-practice (32.3k stars)

Comprehensive Claude Code best practices guide covering CLAUDE.md structure, skill authoring, hook patterns, sub-agent orchestration, and context engineering.

Why it's on this list despite being a guide, not an awesome-list: It functions as a meta-list. Most of the patterns reference specific tools, so reading it gives you a curated tour of the ecosystem with usage context.

Best for: Anyone newer than 60 days into Claude Code who wants the patterns first, the tools second.

Side-by-Side: Lists vs Assembled Systems

Lists You Can Skip

A few lists exist that look comprehensive but waste your time:

• Auto-generated star aggregators that just sort GitHub by query results. Quemsah is the exception because the metrics are useful. Most are not.
• Single-author "my favorite tools" gists with under 50 stars and no recent commits. The signal-to-noise is too low to be worth scanning.
• Lists that haven't shipped a commit in 6+ months. The ecosystem moves too fast for stale lists to be reliable.

Check the last commit date before bookmarking anything not on this page.

How to Actually Use Awesome-Lists

The trap with awesome-lists is treating them as install lists. They are discovery lists. The pattern that works:

1. Bookmark hesreallyhim and one MCP list (wong2 or punkpeye)
2. Reference them when you hit a specific problem (need a Linear MCP, need a debug skill)
3. Install one tool per problem, not five
4. Re-check the lists monthly for new entries in categories you already use

Trying to install everything from any of these lists is the path to context bloat and tool soup. The point of curation is to give you confidence in one pick, not options to chase all of them.

When to Skip the Lists Entirely

Three scenarios where browsing lists is the wrong move:

1. You're in week one of Claude Code. Lists assume you know which problem you're solving. If you don't yet, lists become procrastination disguised as research.
2. You need an end-to-end production stack. Awesome-lists ship discovery, not assembly. ClaudeFast Code Kit ships 18 specialist agents, 167 skill files, and 5 production hooks pre-wired so you skip the assembly step.
3. You need both engineering and marketing systems. Complete Kit bundles Code Kit and Growth Kit at a discount. Browsing 5 awesome-lists to assemble what Complete Kit ships is a multi-week detour.

What to Read Next

• Best Claude Code skills repos for the engineering and general layer
• Best MCP servers list for our own curated MCP picks
• Multi-agent orchestrators compared for the orchestration layer
• Claude Code marketing and GTM skills for the non-engineering lists
• Claude skills guide for writing the skills these lists end up cataloging

If you find yourself bookmarking five awesome-lists in one week, you're scanning, not building. Pick one tool, install it, ship something. Come back to the lists when you actually hit the next problem.