Project NOVA is a multi-agent AI architecture based on n8n and MCP protocol, connecting 25+ specialized agents for intelligent routing and automation across knowledge management, DAW music production, smart home, and development tools. Includes complete Docker containers, n8n workflows, and system prompts for out-of-the-box deployment.
Overview#
Project NOVA is an open-source multi-agent AI architecture that integrates 25+ specialized agents into a unified AI assistant ecosystem through n8n workflow engine and MCP (Model Context Protocol).
Core Architecture#
- Router Agent: Intelligent router that analyzes user intent and matches the most appropriate specialist agent
- n8n Server: Workflow orchestration center managing agent scheduling and task execution
- MCP Servers: Protocol layer handling communication between agents and external applications
- Docker Containers: Isolated runtime environments for each agent
Agent Domain Coverage#
| Domain | Agents |
|---|---|
| Knowledge Management & Notes | TriliumNext, Blinko, BookStack, Outline, SiYuan, Karakeep, Paperless, OnlyOffice, Memos |
| Development & Repository | CLI Server, Forgejo, Gitea, System Search |
| Media & Creative Tools | Ableton Copilot, OBS, Reaper, Reaper QA, YouTube |
| AI & Automation | Flowise, Langfuse, Puppeteer, RAGFlow, Fetch |
| Monitoring & Smart Home | Home Assistant, Prometheus |
Typical Use Cases#
- Smart home control (Home Assistant integration)
- Music production management (Ableton/Reaper DAW control)
- Knowledge base query and retrieval
- Content research and note automation (Fetch → TriliumNext workflow)
- Development documentation sync (Gitea + BookStack)
Deployment#
# Clone repository
git clone https://github.com/PradeepaRW/project-nova.git
cd project-nova
# Start Docker containers
docker-compose up -d
# Access application
# http://localhost:3000
Repository Structure#
project-nova/
├── agents/ # 27 agent definitions
├── mcp-server-dockerfiles/ # 25 MCP server Dockerfiles
├── n8n-workflows/ # 27 n8n workflows
├── prompt-templates/ # Prompt templates
├── reference-guide/ # Reference guides
└── screenshots/ # Architecture diagrams and UI screenshots
Router Agent Response Format#
SELECTED AGENT: [agent_id]
REASON: [Brief explanation]
USER_MESSAGE: [Original user message]
Project Highlights#
- Complete out-of-the-box ecosystem (Dockerfiles, workflows, prompts included)
- Based on standard MCP protocol, easy to extend
- Agent generation templates for custom extensions
- One-click deployment with Docker Compose
- Primary languages: Shell (51.4%), Dockerfile (48.6%)
Contributors#
- Pradeepa Ransamani (PradeepaRW)
- Dujon Walker (dujonwalker)