By dukenicols
Multi-agent coordination system — specialized AI agents collaborate to build software with worktree isolation, structured handoffs, and domain knowledge.
Use when implementing backend APIs, business logic, server-side functionality, middleware, or service layers.
Use when designing database schemas, writing migrations, creating indexes, optimizing queries, or making data model decisions.
Use when implementing UI components, pages, layouts, frontend state management, or client-side functionality.
Use when breaking down requirements into tasks, coordinating specialist agents, managing dependencies, and monitoring progress across the development team.
Use when writing tests (unit, integration, E2E), reviewing code quality, creating test strategies, or validating acceptance criteria.
Delegate a task to the AI development team
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
Initialize AI Development Team for a new project
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin that turns a single AI assistant into a team of specialized agents — backend developer, frontend developer, database engineer, QA engineer — all coordinating through Claude Code's native subagent system.
Instead of one agent context-switching between concerns, you get focused specialists that work in parallel in isolated git worktrees, with structured handoffs and domain knowledge.
You ──> /delegate "Add user auth with OAuth"
│
├─> user-proxy (clarifies requirements, read-only)
│ └─> project-manager (breaks into tasks, coordinates)
│ ├─> backend-developer (APIs, in worktree)
│ ├─> database-engineer (schema, in worktree)
│ ├─> frontend-developer (UI, in worktree)
│ └─> qa-engineer (tests, in worktree)
│
└─> Results reported back with summary
Install once, use in any project. Agents, skills, and hooks are provided by the plugin — only project-specific files (CLAUDE.md, PRD.md, domain docs) live in your repo.
# Add the marketplace
/plugin marketplace add dukenicols/ai-development-team-framework
# Install the plugin
/plugin install ai-dev-team-framework@ai-dev-team
# Enable for your project
/plugin enable ai-dev-team-framework --scope project
Updating:
/plugin update ai-dev-team-framework@ai-dev-team
When the plugin is updated, all your projects get the latest agents, skills, and hooks automatically. Your project-specific files (CLAUDE.md, PRD.md, domain docs) are untouched.
Fork or clone the repo to start a new project from scratch:
git clone https://github.com/dukenicols/ai-development-team-framework.git my-project
cd my-project
claude
With this approach, agents and skills live inside your repo in .claude/agents/ and .claude/skills/. You manage updates manually.
/init-dev-team
Claude will ask about your project name, tech stack, architecture, and domains, then generate:
CLAUDE.md — project configuration with your tech stackPRD.md — product requirements templatecommunication/domains/ — domain knowledge foldersscripts/ — git helper scriptsOpen PRD.md and define your product requirements — vision, target users, phased roadmap, data model, and success metrics.
/delegate Implement user registration with email and password
Each agent is a Markdown file with YAML frontmatter that controls tools, permissions, and isolation:
---
description: "When to use this agent"
tools: [Read, Write, Edit, Bash]
isolation: worktree
maxTurns: 30
---
Specialist agents run in isolated git worktrees — each gets its own branch and working directory. No conflicts when multiple agents work simultaneously.
Subagents return results directly to their parent. No file polling, no message queues. The project-manager aggregates results and returns a summary.
communication/domains/ is the shared state between agents. Each domain folder has business rules, schema decisions, and API contracts that agents read before implementing.
ai-development-team-framework/
├── .claude-plugin/
│ ├── plugin.json # Plugin metadata
│ └── marketplace.json # Marketplace listing
├── agents/ # 6 specialist agents
│ ├── user-proxy.md # Read-only, interprets intent
│ ├── project-manager.md # Coordinates, spawns agents
│ ├── backend-developer.md # Backend, worktree isolated
│ ├── frontend-developer.md # Frontend, worktree isolated
│ ├── database-engineer.md # Database, worktree isolated
│ └── qa-engineer.md # Testing, worktree isolated
├── skills/ # 2 skills (slash commands)
│ ├── delegate/SKILL.md # /delegate — assign work
│ └── init-dev-team/SKILL.md # /init-dev-team — bootstrap
├── hooks/hooks.json # SubagentStop logging
├── settings.json # Permissions allowlist
├── scripts/ # Git helpers (copied to projects)
│ ├── git-agent-commit.sh
│ └── git-agent-merge.sh
├── templates/ # Templates for /init-dev-team
│ ├── CLAUDE.md.template
│ └── PRD.md.template
└── communication/
└── _templates/
└── domain-readme.md # Domain folder template
After running /init-dev-team, your project gets only project-specific files:
npx claudepluginhub dukenicols/ai-development-team --plugin ai-dev-team-frameworkFeature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.