This repository provides a collection of 33 specialized AI subagents designed to enhance Claude Code's capabilities across the software development lifecycle. It targets developers seeking domain-specific expertise and intelligent automation for tasks ranging from frontend development and backend architecture to infrastructure, security, and data science. The benefit is an extended, more capable AI development assistant that can handle complex, multi-domain tasks through intelligent orchestration.

How It Works

The subagents are designed as specialized experts, each with a defined role, expertise, and system prompt. Claude Code intelligently analyzes task context to automatically delegate to the most appropriate subagent(s) or allows explicit invocation. A key component is the agent-organizer, which orchestrates complex, multi-step projects by assembling and managing teams of subagents, ensuring seamless coordination and quality gates. This multi-agent approach aims to provide deeper, more accurate, and comprehensive AI assistance than a single monolithic model.

Quick Start & Requirements

• Installation: Clone the repository to ~/.claude/agents/ or copy individual agent files.
• Prerequisites: Claude Code, Node.js (for MCP servers).
• MCP Server Configuration: Optional but recommended for enhanced performance, requiring configuration in ~/.claude.json with specific Node.js packages like @modelcontextprotocol/server-sequential-thinking and @upstash/context7-mcp.
• Project-Scope Orchestration: For complex projects, copy CLAUDE.md to the project root.
• Documentation: Claude Code Documentation

Highlighted Details

• Features 33 specialized subagents covering frontend, backend, infrastructure, data, AI, security, and QA.
• Includes an agent-organizer for complex multi-agent task coordination and intelligent delegation.
• Supports optional MCP server configurations for enhanced agent capabilities (e.g., sequential thinking, filesystem access).
• Provides detailed examples with token usage and execution time metrics for multi-agent workflows.

Maintenance & Community

The repository is maintained by lst97. There are no explicit mentions of community channels or active contributors beyond the primary author in the README.

Licensing & Compatibility

The repository's license is not explicitly stated in the README. The installation instructions reference a scaffold from wshobson/agents.git, which is under the MIT License.

Limitations & Caveats

The README notes that MCP servers significantly enhance capabilities but are not strictly required for basic functionality. It also advises against placing CLAUDE.md in the global ~/.claude/ directory to avoid over-orchestration of simple tasks, suggesting project-specific usage. Some advanced features may require specific API keys for MCP servers.