claude-devfleet

The claude-devfleet skill provides a robust framework for orchestrating multi-agent coding tasks using Claude DevFleet. It addresses the challenge of managing complex coding projects by enabling users to plan projects, dispatch multiple AI coding agents in parallel, and monitor their progress in a structured and automated way. Each agent operates within an isolated git worktree, ensuring clean environments and avoiding interference between concurrent tasks. The skill integrates seamlessly with a running Claude DevFleet instance via MCP, providing a streamlined approach to collaborative AI-driven software development.

Core features include automated project planning, mission creation, and intelligent task dispatch. Users can generate a project plan from a high-level prompt, breaking it down into missions with dependency chains. Agents can be dispatched automatically or manually, with real-time monitoring through mission status checks or dashboard overviews. Structured reports capture changes, errors, and next steps, enabling informed decision-making and facilitating iterative development. Concurrency is managed efficiently, with a default of three simultaneous agents, ensuring that task execution scales without overwhelming system resources.

This skill is ideal for software engineers, project managers, and AI enthusiasts who need to coordinate multiple coding workflows efficiently. It supports both fully automated pipelines, where tasks flow seamlessly from planning to reporting, and manual step-by-step control for fine-grained oversight. Typical use cases include building APIs, implementing features across multiple modules, running automated tests, and tracking project progress across multiple interdependent tasks. By automating task orchestration, claude-devfleet reduces manual overhead, improves consistency, and accelerates the development lifecycle.