hc

🤖 HyperClaude Nano Plugin

An official Claude Code plugin that transforms your AI assistant into a coordinated team of specialized agents with intelligent task orchestration, advanced MCP integration, and professional-grade quality.

✨ What Is HyperClaude Nano?

HyperClaude Nano is an official Claude Code plugin that turns your AI assistant into an intelligent, coordinated team of specialist agents. Think of it as giving each developer their own personal development staff - with architects, coders, designers, security experts, testers, technical writers, and cloud engineers working together seamlessly.

Core Architecture

🧠 Intelligent Agent System

• 7 Specialized Agents - Domain experts with specific tools and capabilities
• Wave Orchestration - Multi-phase execution for complex projects (W1→W2→W3→W4→W5)
• Reference-Based Communication - Agents share findings via structured protocols
• MCP-Powered Intelligence - 5 integrated servers providing superpowers

⚡ Advanced Features

• Mandatory Task Tracking - TodoWrite enforced for 3+ step operations
• Pattern-Aware Analysis - Respects existing codebase conventions
• Parallel Execution - Multiple agents work simultaneously when possible
• Quality Gates - Built-in validation at every step
• Context Optimization - 40-50% token reduction via intelligent caching

How It Works

Intelligent Routing: The framework analyzes your request and automatically selects the optimal agent(s) and execution strategy.

Wave Execution: For complex tasks, agents work in coordinated phases:

1. W1 (Architect) - System analysis and planning
2. W2 (Security) - Vulnerability assessment
3. W3 (Coder + Designer) - Parallel implementation
4. W4 (Test Engineer) - Quality validation
5. W5 (Tech Writer) - Documentation

Memory Persistence: All patterns, decisions, and findings are stored in the MCP memory server for cross-session intelligence.

Quality Assurance: Every operation goes through mandatory validation gates with evidence-based completion requirements.

🏗️ Architecture

Core Systems

🤖 Agent Orchestration

• 7 Specialized Agents - Each with distinct tools, capabilities, and domain expertise
• Task Tool Integration - Agents coordinate via Claude Code's Task system
• Reference-Based Communication - Structured handoff protocols (see plugin/AGENT_PROTOCOLS.md)
• Memory Persistence - Shared knowledge via MCP memory server

📋 Task Management System

• Mandatory TodoWrite - Enforced for operations with 3+ steps
• Status Tracking - pending → in_progress → completed with evidence requirements
• Quality Gates - Cannot mark complete without validation
• Progress Visibility - Real-time task status for users

🌊 Wave Orchestration

• Trigger Conditions - >15 files, >5 types, >3 domains, or "comprehensive" requests
• Coordinated Phases - Sequential and parallel execution strategies
• Dependency Management - Smart ordering of agent activation
• Shared Context - Each wave builds on previous findings

MCP Server Integration

The plugin leverages 5 MCP servers for enhanced capabilities:

• 🧠 Memory Server - Cross-session pattern storage, agent communication, knowledge persistence
• 🔍 Tree-Sitter Server - AST analysis, code structure understanding, pattern detection
• 📚 Context7 Server - Live documentation lookup, framework best practices, library APIs
• 🌐 Puppeteer Server - Visual testing, screenshot generation, E2E automation
• 🔢 Sequential-Thinking Server - Complex problem decomposition, multi-step reasoning

Automatic Configuration

The plugin's .mcp.json file contains all MCP server configurations. When you install the plugin, these servers are automatically available - no additional setup needed!

Optimization Benefits:

• Token reduction via Memory caching
• Faster analysis via Tree-Sitter
• More efficient documentation lookup via Context7
• Parallel operation capabilities
• Structured data persistence

🤖 Agent Specifications

Each agent is a domain specialist with specific tools, capabilities, and coordination protocols:

🏗️ Architect Agent

System Architecture Analysis & Pattern Identification