This repository provides a collection of 56 specialized AI subagents designed to enhance development workflows within Claude Code. These agents offer domain-specific expertise, automatically invoked based on context or explicitly called, to improve efficiency and performance across various development, operations, and business tasks.

How It Works

The subagents are structured as individual files, each defining a specific role, invocation criteria, and optionally, a preferred Claude model (Haiku, Sonnet, or Opus) based on task complexity. This allows for intelligent delegation, using cost-effective models for simpler tasks and more powerful ones for critical operations. Claude Code orchestrates these agents through sequential, parallel, conditional, and review-based workflows, enabling complex multi-agent task execution.

Quick Start & Requirements

• Installation: Clone the repository into ~/.claude/agents/.

  cd ~/.claude
  git clone https://github.com/wshobson/agents.git
  

• Usage: Agents are invoked automatically by Claude Code based on context or explicitly by name (e.g., "Use the code-reviewer...").
• Documentation: Claude Code Documentation

Highlighted Details

• 56 Specialized Agents: Covers a broad spectrum from backend/frontend development, language specialists (Python, Go, Rust, C++, Java, etc.), DevOps, security, data science, ML, and business functions.
• Model Tiering: Agents are configured to use Claude Haiku (cost-effective), Sonnet (balanced), or Opus (maximum capability) based on task complexity.
• Multi-Agent Workflows: Supports complex orchestrations like feature development, incident response, and ML pipeline creation, with examples provided.
• Explicit Invocation: Allows users to directly target specific agents for precise control over task delegation.

Maintenance & Community

The repository is maintained by wshobson. Further community interaction details are not specified in the README.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive license suitable for commercial use and integration with closed-source projects.

Limitations & Caveats

The effectiveness of automatic invocation relies heavily on clear and contextually rich user requests. Conflicting advice between agents is possible and may require explicit user intervention for reconciliation. The README does not detail specific versioning or release management practices.