// open_source
Claude
Battle-tested Claude Code global configuration with specialized agents, slash commands, auto-triggered skills, and MCP server definitions for AI-assisted development
View on GitHubThis is my personal global configuration for Claude Code, providing a structured setup for agents, commands, skills, and global settings that apply across all my projects.
Motivation
Claude Code becomes significantly more capable with well-organized configuration. This repository captures my battle-tested setup for orchestrating AI-assisted development workflows, from .NET backend work to infrastructure-as-code to frontend & mobile development.
Structure
.claude/
├── CLAUDE.md # Global instructions (orchestrator behavior, style)
├── settings.json # Permissions, plugins, and settings
├── mcp-servers.json # MCP server definitions
├── agents/ # Specialized sub-agent definitions
├── commands/ # Slash command definitions
└── skills/ # Auto-triggered skills based on context
Agents
Specialized sub-agents invoked for domain-specific work:
| Agent | Purpose |
|---|---|
| dotnet | .NET 10 backend, Goa framework, Lambda-first, vertical slices |
| iac | AWS infrastructure, OpenTofu, IAM, security |
| web | Next.js App Router, Tailwind, accessibility |
| uiux | UX critique, interaction design, cognitive load |
| copywriting | Headlines, landing pages, developer messaging |
Skills
These context-aware capabilities trigger automatically when relevant, including .NET source generator configuration, AOT analysis, package management, and more. Many skills are sourced from the companion dotnet-skills repository.
MCP Servers
Includes configuration for sequential thinking, Playwright browser automation, and AWS documentation and knowledge servers. Setup instructions are provided for both npx and uvx based servers.