MCP Sub-Agents for Enterprise Development¶

A comprehensive collection of specialized Claude Code agents with domain expert savants and hybrid workflow commands for building production-ready, enterprise-grade applications.

Overview¶

MCP Sub-Agents implements a two-tier agent architecture designed to provide deep technical expertise across multiple technology stacks. This system combines the broad knowledge of savant orchestrators with the focused expertise of specialist implementers.

graph TB
    subgraph tier1["TIER 1: SAVANTS"]
        js[fullstack-js-savant<br/>JavaScript/TypeScript Master]
        java[java-spring-savant<br/>Spring Ecosystem Master]
    end

    subgraph tier2["TIER 2: SPECIALISTS"]
        subgraph jsstack[JavaScript Stack]
            jsspec1[fullstack-nextjs]
            jsspec2[frontend-ui]
            jsspec3[backend-api]
        end

        subgraph crosscut[Cross-Cutting]
            cross1[code-reviewer]
            cross2[qa-tester]
            cross3[security]
            cross4[performance]
            cross5[deployment]
            cross6[database]
            cross7[monitoring]
            cross8[architect]
            cross9[tech-writer]
        end
    end

    tier1 -->|Delegates to| tier2

Key Features¶

Two-Tier Architecture¶

Tier 1: Savant Agents

Master orchestrators with broad, deep knowledge across entire technology ecosystems. Savants make architectural decisions, choose technologies, and coordinate specialist agents.

fullstack-js-savant - JavaScript/TypeScript ecosystem master
java-spring-savant - Spring ecosystem master

Tier 2: Specialist Agents

Deep domain experts focusing on specific frameworks, tools, and practices within a stack.

JavaScript specialists for Next.js, UI, and APIs
Cross-cutting specialists for security, testing, performance, and more

Hybrid Workflow Commands¶

Structured slash commands that combine development workflow stages with technology expertise for guided, structured development:

/mcp-design-architecture - System architecture design
/mcp-implement-fullstack - End-to-end feature implementation
/mcp-review-code - Comprehensive code quality review
/mcp-qa-e2e - End-to-end testing with Playwright
/mcp-deploy - Production deployment with CI/CD

View all workflow commands

Why Use MCP Sub-Agents?¶

Production-Ready Expertise¶

Each agent provides comprehensive guidance for building enterprise-grade applications with modern best practices, security considerations, and performance optimizations.

Flexible Usage¶

Use savants for complex architectural decisions or specialists for focused implementation tasks. Mix and match based on your needs.

Guided Workflows¶

Workflow commands provide step-by-step guidance through design, implementation, review, QA, and deployment phases.

Technology-Specific¶

Deep expertise in modern JavaScript/TypeScript stack including Next.js 15+, React Server Components, tRPC, Prisma, and more.

Quick Start¶

Installation¶

Install all agents and workflow commands with one command:

curl -fsSL https://raw.githubusercontent.com/rcdelacruz/mcp-sub-agents/main/install.sh | bash

View detailed installation instructions

First Steps¶

Use a savant for architecture decisions:

Use fullstack-js-savant to design a blog application architecture

Use specialists for implementation:

Use fullstack-nextjs to build the blog post page with Server Components
Use frontend-ui to create the blog post form with shadcn/ui

Use workflow commands for guided development:

/mcp-design-nextjs
/mcp-implement-fullstack
/mcp-review-code

View complete quick start guide

Architecture¶

The two-tier architecture provides both breadth and depth:

flowchart LR
    User[Developer] --> Choice{Task Type}

    Choice -->|Complex/Architecture| Savant[Savant Agent]
    Choice -->|Specific/Implementation| Specialist[Specialist Agent]

    Savant --> Design[Architectural Design]
    Savant --> Orchestrate[Orchestrate Specialists]

    Specialist --> Implement[Deep Implementation]

    Design --> Result[Complete Solution]
    Orchestrate --> Result
    Implement --> Result

When to Use Savants¶

Architectural decisions and system design
Technology stack selection
Complex cross-cutting problems
Full-stack feature orchestration
High-level technical guidance

When to Use Specialists¶

Framework-specific implementation
UI component development
API endpoint creation
Database schema design
Security audits
Performance optimization

Learn more about the architecture

Technology Stack¶

MCP Sub-Agents provides expertise in modern JavaScript/TypeScript development:

Frontend¶

Next.js 15+ with App Router
React 18+ with Server Components
TypeScript
Tailwind CSS and shadcn/ui
Radix UI primitives

Backend¶

Next.js API Routes and Server Actions
tRPC for type-safe APIs
NextAuth.js v5 for authentication
Node.js and Express/NestJS

Database¶

PostgreSQL with Prisma ORM
Redis for caching
Database optimization and migrations

Testing & Quality¶

Playwright for E2E testing
Vitest for unit testing
ESLint and Prettier
Code review best practices

DevOps¶

Vercel deployment
Docker containerization
GitHub Actions CI/CD
Monitoring with Sentry

View complete tech stack reference

Example Workflows¶

Building a Blog Feature¶

# 1. Design the architecture
/mcp-design-nextjs

# 2. Implement end-to-end
/mcp-implement-fullstack

# 3. Review code quality
/mcp-review-code

# 4. Security audit
/mcp-review-security

# 5. Write tests
/mcp-qa-e2e

# 6. Deploy to production
/mcp-deploy

Creating an Authentication System¶

# Use savant for architecture
Use fullstack-js-savant to design authentication with NextAuth.js

# Use specialist for implementation
Use backend-api to implement NextAuth.js with GitHub OAuth

# Review security
/mcp-review-security