A Claude Code skill that makes Claude an expert at configuring Claude Code
A meta-skill that Claude Code uses to expertly guide you through creating skills, subagents, hooks, permissions, and MCP server integrations. When installed, Claude becomes your configuration expert.
Open Claude Code and run these commands:
/plugin marketplace add werkamsus/claude-metaskill
/plugin install claude-code-config
Done. Claude now knows how to configure itself.
When installed, Claude Code automatically uses this skill to provide expert guidance on:
- ✨ Creating Skills - File structure, SKILL.md format, best practices
- 🤖 Building Subagents - YAML configuration, tool permissions, model selection
- 🪝 Configuring Hooks - Auto-format on save, notifications, validation
- 🔐 Managing Permissions - Allow/ask/deny patterns, security best practices
- 🔌 Setting Up MCP Servers - Context7, Playwright, Postgres integration
- 📚 Real-World Examples - Production-ready configurations and patterns
- 🔧 Troubleshooting - Common issues and solutions
- Visit SkillsMP.com
- Search for "claude-metaskill"
- Click install and follow instructions
# Add the marketplace
/plugin marketplace add werkamsus/claude-metaskill
# Install the skill
/plugin install claude-code-configGlobal (all projects):
git clone https://github.com/werkamsus/claude-metaskill.git
cp -r claude-metaskill/skills/claude-code-config ~/.claude/skills/Project-specific:
git clone https://github.com/werkamsus/claude-metaskill.git
cp -r claude-metaskill/skills/claude-code-config .claude/skills/Once installed, the skill activates automatically when you ask about Claude Code configuration:
"How do I create a new skill?"
"Set up a subagent for code review"
"Configure Context7 MCP"
"Add a hook to auto-format TypeScript files"
"Set up permissions to block .env files"
Or explicitly invoke it:
"Use the claude-code-config skill to help me create a test generator subagent"
- Complete configuration reference
- Directory structure and file formats
- Step-by-step instructions
- Quick reference tables
- Troubleshooting section
- Real-world skills (PDF processor, API docs generator, database migrations)
- Production subagents (security auditor, performance optimizer, test generator)
- Advanced hook configurations (auto-format, validation, notifications)
- Complex permission patterns (dev, production, read-only modes)
- MCP server setups (full stack, API development)
- Project-specific configs (Next.js, Django)
You: "I want to create a skill for processing markdown files"
Claude: [Uses claude-code-config skill]
Let me help you create a markdown processing skill.
First, I'll create the directory structure:
~/.claude/skills/markdown-processor/SKILL.md
[Creates skill with proper YAML frontmatter and instructions]
You: "Help me set up Context7 to get up-to-date library docs"
Claude: [Uses claude-code-config skill]
I'll help you install and configure Context7 MCP server.
Here are the steps:
1. Add to your settings.json...
2. Get an API key from context7.com...
3. Configure permissions...
You: "Create a subagent that reviews code for security issues"
Claude: [Uses claude-code-config skill]
I'll create a security-focused code review subagent.
[Creates ~/.claude/agents/security-reviewer.md with:
- Proper YAML configuration
- Security checklist
- Tool restrictions
- Clear instructions]
claude-metaskill/
├── skills/claude-code-config/
│ ├── SKILL.md # Main reference guide
│ └── examples.md # Real-world examples
├── .claude-plugin/
│ └── marketplace.json # Plugin marketplace config
├── README.md # This file
└── LICENSE # MIT License
Found an issue or want to add examples? Contributions welcome!
- Fork the repository
- Create your feature branch
- Add your examples or improvements
- Submit a pull request
- Skills: Creation, structure, best practices, activation triggers
- Subagents: YAML frontmatter, tool permissions, model selection
- Hooks: PreToolUse, PostToolUse, Stop, Notification events
- Permissions: Allow/ask/deny patterns, security configurations
- MCP Servers: Context7, Playwright, Postgres, custom servers
- Settings: Hierarchy, precedence, environment variables
- Commands: /config, /agents, /plugin management
MIT License - feel free to use, modify, and share!
Created by Nick Raziborsky
Built with knowledge from:
- Anthropic's official Claude Code documentation
- Community best practices
- Real-world configuration patterns
Make Claude Code truly yours - master configuration with Claude MetaSkill ✨