Does CDT require any experimental features?

Yes, it requires Claude Code with the CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS environment variable set to 1 to enable the teammate delegation functionality.

How does the lead agent interact with the teammates?

The lead agent acts as a coordinator, delegating tasks to the Developer or Architect and relaying research findings, while the teammates communicate directly with each other to resolve bugs and reviews.

What is the main benefit of using CDT over standard Claude Code?

CDT introduces specialized roles like Architects and Reviewers who debate and validate work, ensuring higher quality and better adherence to project architecture than a single agent could achieve.

Can I use CDT for just planning without writing code?

Absolutely. The 'plan' mode is specifically designed for the Architect and PM teammates to debate trade-offs and create a comprehensive plan.md without modifying source files.

Claude Dev Team (CDT)

Name: Claude Dev Team (CDT)
Author: rube-de

byrube-de

•

Herramientas de Colaboración

Orchestrates a multi-agent AI development team to architect, implement, and review complex software projects with automated quality gates.

Claude Dev Team (CDT) transforms Claude Code into a comprehensive multi-agent workspace by delegating tasks across specialized roles such as Architects, Product Managers, and QA Testers. By utilizing a structured workflow with four modes—ranging from strategic planning to fully automated execution—CDT ensures high-quality output through peer-to-peer messaging between agents, mandatory code reviews, and integrated testing cycles. It is the ideal solution for developers managing complex features where architectural integrity, rigorous verification, and collaborative problem-solving are essential.

Características Principales

01Specialized agent roles including Architect, Developer, Researcher, and Reviewer teammates.

025 GitHub stars

03Four distinct operational modes (Plan, Dev, Full, and Auto) for flexible workflow management.

04Peer-to-peer agent messaging for collaborative debate and peer-driven quality control.

05Built-in Architectural Decision Record (ADR) management to maintain long-term project integrity.

06Automated quality gates featuring mandatory unit, integration, and UI testing via specialized testers.

Casos de Uso

01End-to-end feature development from initial architectural design to final production-ready implementation.

02Large-scale codebase refactoring where planning and cross-functional validation are required before execution.

03Automated code reviews and security audits using a multi-agent 'AI Council' approach.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rube-de/cc-skills cdt

For use in Claude.ai and ChatGPT

Download Skill