Awesome Claude Code: Best Skills, Tools, and Community Resources

The official Claude Code repository crossed 131,000 GitHub stars by June 2026, making it one of the fastest-growing developer tools in GitHub history. But the real story is what happened around it. Community-maintained awesome lists, skills repositories, and companion tools have collectively gathered over 200,000 stars, with the largest single aggregator (affaan-m/everything-claude-code) passing 163,000 stars on its own by mid-2026.

That growth reflects a structural shift in how developers extend their tools. Claude Code's extension model relies on three primitives: skills (Markdown instruction packs loaded on demand), hooks (shell commands that fire on specific events), and MCP servers (external programs that expose tools via the Model Context Protocol). None of these require a package registry or app store. A skill is a folder with a SKILL.md file. A hook is a JSON entry in settings.json. An MCP server is a URL you point Claude Code at. This low-friction model means anyone can publish an extension by pushing a GitHub repo, and the awesome lists became the discovery layer that GitHub's own search doesn't provide.

The practical problem is volume. With 500-plus resources scattered across a dozen lists, finding the right tool for a specific workflow takes more time than it should. The rest of this guide organizes the landscape by what each list covers and which tools are worth installing first.

Not all awesome lists serve the same purpose. Some are hand-curated with editorial judgment. Others are automated aggregators that index everything. Here are the lists worth knowing, organized by what they're best at.

1. hesreallyhim/awesome-claude-code (47.6k stars)

The most widely referenced list. It covers skills, hooks, slash commands, agent orchestrators, applications, and plugins with consistent editorial judgment. Tools that don't work get removed. This is the list to start with if you want one bookmark.

2. affaan-m/everything-claude-code (163k+ stars)

The largest aggregator by volume and star count. Originally called "Everything Claude Code," it functions as a discovery firehose rather than a curated recommendation. Good for finding niche tools, but expect to filter heavily. The project has since evolved into ECC, an agent harness optimization system with skills, instincts, memory, and security features that work across Claude Code, Codex, and Cursor.

3. rohitg00/awesome-claude-code-toolkit (2.2k stars)

The most structured list, organized into 10 agent categories with exact counts: 135 agents, 35 skills, 42 commands, 176+ plugins, 20 hooks, 15 rules, 7 templates, and 14 MCP configurations. Useful when you know you need an agent for a specific domain (infrastructure, QA, data science) and want to compare options within that category.

4. jqueryscript/awesome-claude-code (446 stars)

A broad catalog with 500+ entries organized by function: agents and orchestration, IDE integrations, clients and GUIs, infrastructure and proxies, usage and observability. Updated frequently (last update June 26, 2026). Best for finding tools in categories the larger lists don't break out, like alternative Claude Code clients and usage analytics.

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

More guide than list. This repository combines curated tool recommendations with patterns and context for how to use them effectively. Start here if you want opinionated workflow advice alongside your tool discovery.

Give Your Claude Code Agents a Persistent Workspace

Fast.io connects to Claude Code as an MCP server, giving your agents versioned file storage, built-in semantic search, and human-agent handoff in one workspace. Starts with a 14-day free trial.

Start 14-Day Trial

Before diving into specific tools, it helps to understand the three extension types and when to use each one.

Skills are folders containing a SKILL.md file and optional helper scripts. Claude Code loads them on demand when you invoke them as slash commands. Think of a skill as a reusable prompt template that can also include scripts, validation logic, or data files. Install one by dropping a folder into ~/.claude/skills/ (global) or .claude/skills/ (project-specific).

A good example: a /review skill that defines your team's code review checklist, invoked with a single command instead of pasting the same 200-word prompt each time.

Hooks are shell commands that fire automatically on specific events: session start, tool use, prompt submission, or context compaction. You configure them in settings.json. Hooks enforce rules without manual intervention. A pre-commit hook might run a linter before Claude Code creates a commit. A session-start hook might load project context from a database.

MCP servers are external programs that speak the Model Context Protocol, giving Claude Code access to systems it can't reach natively. A PostgreSQL MCP server lets Claude Code query your database. A GitHub MCP server lets it read issues and PRs with full context. A Slack MCP server lets it search team conversations.

The recommended setup pattern from the community: start with one scoped .mcp.json containing only the MCP servers you actually use, one safety hook (like a pre-commit linter), and one reusable skill for your most repeated workflow. Add more as specific needs arise, not all at once.

Across all the awesome lists, certain tools consistently appear at the top. Here are the standouts in each category, verified against their GitHub repositories.

Agent Orchestration -

Claude-Flow (59.4k stars): A code-first orchestration layer for running multi-agent workflows. Handles task delegation, parallel execution, and result aggregation across multiple Claude Code instances

• gstack (116.2k stars): Garry Tan's opinionated setup that assigns specialized agent roles (CEO, Engineering Manager, Release Manager, QA Engineer) for structured product development workflows
• oh-my-claudecode (9.9k stars): Teams-first orchestration with 19 specialized agents and 28 skills, designed for collaborative development environments

Skills and Knowledge Packs

Superpowers (227.6k stars): A comprehensive skills library covering software engineering patterns. Provides proven techniques and development patterns that Claude Code can invoke on demand

• Claude Scientific Skills (28.1k stars): Ready-to-use agent skills for research, science, engineering, analysis, finance, and technical writing
• ComposioHQ/awesome-claude-skills: Focused specifically on SaaS workflow automation skills, useful if your work involves connecting Claude Code to business tools

Usage and Observability

• ccusage (11.5k stars): CLI tool that analyzes Claude Code usage from local files, generating daily, monthly, and session-level reports. Useful for tracking token spend and understanding usage patterns
• ccpm (7.6k stars): Project management using GitHub Issues and git worktrees for parallel agent execution. Tracks what each Claude Code session is working on across branches

Developer Workflow

• pro-workflow (1.8k stars): Battle-tested workflows from power users, including self-correcting memory, parallel worktree patterns, and wrap-up rituals
• claude-mem (35.9k stars): Automatically captures and compresses Claude Code activity, then injects relevant context into future sessions. Addresses the persistent-memory gap in long-running projects
• claude-code-hooks-mastery (3.3k stars): A complete guide to hooks with UV scripts and meta-agents. Good starting point if hooks are new to you

Quality and Safety

Bouncer: An independent quality gate that uses a second LLM (Gemini) to audit Claude Code's output, triggered automatically through hooks

• mcpick (by spences10): A Claude Code extension manager for MCP servers, plugins, and skills. Simplifies installation and configuration instead of manual JSON editing

One gap the awesome lists surface repeatedly is persistent storage. Claude Code sessions are ephemeral. Files created during a session live on your local machine, and if you're running agents in CI, containers, or cloud environments, those files disappear when the session ends.

Fast.io solves this as an intelligent workspace where agents and humans share the same files, permissions, and search layer. Here is what that looks like in practice:

• MCP-native access: Fast.io exposes a Streamable HTTP endpoint at /mcp and legacy SSE at /sse. Claude Code connects to it like any other MCP server, gaining access to a consolidated MCP toolset for workspace, storage, AI, and workflow operations
• Built-in RAG: Enable Intelligence Mode on a workspace, and every uploaded file gets auto-indexed for semantic search. No separate vector database, no embedding pipeline. Ask questions about your files and get answers with citations
• Ownership transfer: An agent creates workspaces, uploads files, builds shares, then transfers ownership to a human. The agent keeps admin access for ongoing maintenance, and the human gets a ready-to-use workspace
• File versioning and locks: Every file change is versioned. Acquire locks to prevent conflicts when multiple agents write to the same workspace

For teams using Claude Code alongside other tools like local storage, S3, or Google Drive, Fast.io adds the collaboration and intelligence layer those options lack. Local storage doesn't sync across machines. S3 doesn't index files for semantic search. Google Drive doesn't expose an MCP server. Fast.io does all three, and every org starts with a 14-day free trial.

You can connect Claude Code to Fast.io in one command:

claude mcp add fastio --transport http /storage-for-agents/

The most common mistake with awesome lists is installing everything that looks interesting. Twenty MCP servers and fifty skills create more noise than signal. Claude Code's tool selection degrades past about 50 visible tools, and excess context from unused skills wastes tokens on every interaction.

Here is a practical setup sequence based on what the community has converged on:

Week 1: Foundation

Start with a short CLAUDE.md file in your project root that describes your codebase conventions, tech stack, and testing approach. Add one MCP server for your most-used external system (GitHub, your database, or your project tracker). Write one skill for the workflow you repeat most often, whether that's code review, deployment, or documentation.

Week 2: Safety and Observability

Add a pre-commit hook that runs your linter before Claude Code creates commits. Install ccusage or a similar analytics tool to understand your token spend. Set up one quality-gate hook if you work on a codebase where errors are expensive.

Week 3: Team Patterns

If you work with a team, evaluate an orchestration tool like Claude-Flow or oh-my-claudecode for multi-agent workflows. Share your CLAUDE.md and skills as a .claude/ directory in your repository so the whole team gets the same setup.

Ongoing: Audit and Prune

Every month, check which skills and MCP servers you actually used. Remove the ones you didn't. The awesome lists will still be there when you need something new.

For persistent storage across all these workflows, Fast.io's MCP server gives Claude Code a workspace where files, search, and permissions survive between sessions. The skill documentation covers the full tool surface for workspace and storage operations.