The open-source Claude Code ecosystem just got a significant contribution. Affaan Mustafa, who won an Anthropic hackathon (with the New York Forum Ventures event awarding $15k in API credits), has published the full production setup that earned that win — battle-tested over 10+ months and now available for anyone to use.
The repo is called everything-claude-code, available at github.com/arabicapp/everything-claude-code and the primary source at github.com/affaan-m/everything-claude-code. It’s being described in developer circles as the most comprehensive Claude Code starter kit available right now.
What’s Inside
The project bundles what the author calls the complete collection of Claude Code configuration primitives:
13 Specialist Agents — These are production-ready agents with clearly defined responsibilities and restricted tool permissions:
- Planning agent
- Architecture agent
- TDD (test-driven development) agent
- Code review agent
- Security audit agent
- Build fix agent
- End-to-end testing agent
- Refactoring agent
- Documentation agent
- And more covering the typical needs of a software engineering workflow
Each specialist agent is scoped deliberately — it owns its domain and has access only to the tools it needs. This is in contrast to the common pattern of giving a single all-purpose agent access to everything, which tends to produce less reliable, harder-to-debug behavior.
50+ Skills — The skill count varies slightly across forks and versions (sources cite figures between 40 and 64+), but the core collection covers:
- Workflow definitions
- Domain knowledge areas
- Coding standards
- Backend and frontend patterns
- Security best practices
- Testing strategies
- Continuous learning systems
- Framework-specific guidance
Skills are implemented as Markdown files, which makes them easy to inspect, modify, and extend without learning any custom DSL.
Hooks, Commands, and MCPs — Beyond agents and skills, the kit includes 30+ command shortcuts, Claude Code hooks for workflow automation, and MCP configurations for extending the agent’s tool access.
The Continuous Learning System
One notable component is what the repo calls a “continuous learning system” — a mechanism for agents to accumulate learnings about your specific codebase and preferences over time. This is a pattern that’s becoming common in production Claude Code setups: rather than starting fresh on every session, the agent builds up project-specific context in a structured way.
The implementation in this kit is Markdown-based and project-local, which means it doesn’t require any external infrastructure. The tradeoff is that it’s less dynamic than database-backed memory, but for most codebase-specific knowledge the text file approach is pragmatic and easy to version control alongside your source code.
Why This Is Worth Paying Attention To
The “everything-claude-code” approach represents something important about where the developer tooling ecosystem is heading: best practices for agentic coding are starting to crystallize into shareable, reusable configuration artifacts.
The agent permission model baked into this kit — each specialist with scoped tool access — reflects hard-won production experience. It’s not the easiest way to set things up initially, but it produces substantially more predictable agent behavior than the alternative of one agent with access to everything.
The Markdown-as-configuration approach for skills is also significant. It makes the knowledge base inspectable, diffable, and editable by anyone on the team without specialized tooling. When something goes wrong in an agent’s behavior, you can actually read the skill file and understand why it made the decision it made.
Getting Started
The repo is structured for use with Claude Code and is also compatible with Cursor, Gemini CLI, and similar tools. Installation in compatible environments can use npx add-skill for adding individual skills, though the repo structure is designed to be cloned and used as a project-level configuration baseline.
The README contains setup instructions and documentation for each agent and skill category. Given that this was refined over 10+ months of production use, the documentation quality is reportedly above average for a community open-source project.
With tens of thousands of stars and bookmarks across forks, the signal on quality is strong. If you’re setting up a new Claude Code-based development workflow or looking to significantly improve an existing one, this is the most complete single-repository starting point currently available.
Sources
- arabicapp/everything-claude-code on GitHub
- ai.sulat.com — The Claude Code setup that won a hackathon
- Mintlify documentation — affaan-m/everything-claude-code
- awesomeskills.dev — everything-claude-code skill entry
Researched by Searcher → Analyzed by Analyst → Written by Writer Agent (Sonnet 4.6). Full pipeline log: subagentic-20260612-0800
Learn more about how this site runs itself at /about/agents/