🚀 FlowTech-AI: Production AI Stack for Developers

Self-hosted AI infrastructure: Cursor AI assistant + Conversational AI (OpenWebUI) + RAG + Intelligent Automation (n8n)

---

🎯 What is FlowTech-AI?

A production-ready, self-hosted AI stack that combines:

• 🤖 AI Services: OpenWebUI interface, Qdrant vector DB, n8n automation, Ollama LLM
• 💻 Cursor Integration: MCP-Qdrant for AI-enhanced coding with context awareness
• 📚 Knowledge Base: RAG with document upload in OpenWebUI
• 🔄 Intelligent Workflows: n8n automation platform
• 🚀 One-command deployment: ./init.sh and everything works

✨ Key Features

✅ Cursor AI Enhancement via Model Context Protocol (MCP)
✅ OpenWebUI with RAG for document Q&A
✅ Vector Search with Qdrant (1024-dim embeddings)
✅ Production-ready Docker Compose stack
✅ Automated Workflows with n8n orchestration
✅ Fork-friendly - Clone once, everything works

---

🚀 Quick Start (5 minutes)

# 1. Clone repository
git clone https://github.com/FlowTech-Lab/FlowTech-AI.git
cd FlowTech-AI

# 2. Initialize stack (requires sudo for permissions)
sudo ./init.sh

# ✅ Stack ready! Services available at:
# - OpenWebUI: http://localhost:8081
# - Cursor MCP: http://localhost:8000
# - n8n: http://localhost:5678
# - Qdrant: http://localhost:6333

Full guide: See QUICKSTART.md

---

📦 What's Included?

Core Services

Service	Port	Description	Status
OpenWebUI	8081	AI chat interface with RAG	✅ Production
MCP-Qdrant	8000	Cursor code context (cursor-context)	✅ Production
MCP-Knowledge	8001	Cursor notes search (cursor-knowledge)	✅ Production
n8n	5678	Workflow automation	✅ Production
Qdrant	6333	Vector database	✅ Production
Samba	445	Notes share (SMB)	✅ Production
PostgreSQL	5432	Metadata storage	✅ Production
Redis	6379	Cache & queues	✅ Production
SearxNG	8082	Web search engine	✅ Production
Langfuse	3300	LLM observability	✅ Production

---

💻 Cursor Integration

Setup (2 minutes)

# 1. Copy MCP config template
cp cursor-mcp-config.json ~/.cursor/mcp.json

# 2. Edit IP (change to your server IP)
nano ~/.cursor/mcp.json
# Replace 192.168.0.246 with your actual IP

# 3. Restart Cursor

# 4. Test
@qdrant store "FlowTech-AI is awesome!"
@qdrant find awesome

What you get:

• Store code snippets, notes, and context in Qdrant
• Retrieve information semantically during coding
• AI-enhanced development with persistent memory

Sync Your Notes Automatically

Sync Markdown notes to Cursor:

# Initial sync
./scripts/sync-notes.sh

# Install hourly auto-sync
./scripts/install-cron.sh

Your Notes/ folder will be automatically synced to Qdrant and searchable in Cursor!

📖 See Notes Sync Guide for details.

Access Notes via Network Share

Edit notes from Windows/Mac/Linux:

# Windows
\\YOUR_SERVER_IP\notes

# Mac/Linux
smb://YOUR_SERVER_IP/notes

# Credentials (generated in .env during init.sh)
Username: admin
Password: Check your .env file (SAMBA_PASSWORD)

Open the share with Obsidian or any text editor to manage your notes!

📖 See Samba Windows Guide for connection help.

---

📚 OpenWebUI RAG

Upload Documents

1. Open http://localhost:8081
2. Create a new chat
3. Click "Knowledge" button
4. Upload your documents (PDF, MD, TXT, DOCX, etc.)
5. Ask questions about your documents!

Example:

User: What does the API documentation say about authentication?
AI: Based on the uploaded API docs, authentication uses JWT tokens...

Supported Formats

• 📄 PDF, DOCX, TXT, MD
• 💻 Code files (PY, JS, TS, etc.)
• 🌐 Websites (via URL)

---

🔧 Architecture

┌─────────────────────────────────────────────────────────┐
│                   FlowTech-AI Stack                     │
├─────────────────────────────────────────────────────────┤
│                                                         │
│  ┌──────────────┐    ┌──────────────┐    ┌──────────┐ │
│  │  Cursor IDE  │───▶│ MCP-Qdrant   │───▶│  Qdrant  │ │
│  │  (Dev Tool)  │    │  (Port 8000) │    │  Vector  │ │
│  └──────────────┘    └──────────────┘    │    DB    │ │
│                                           └────┬─────┘ │
│  ┌──────────────┐    ┌──────────────┐         │       │
│  │   Browser    │───▶│  OpenWebUI   │────────▶│       │
│  │              │    │  (Port 8081) │         │       │
│  └──────────────┘    └──────────────┘         │       │
│                                                │       │
│  ┌──────────────┐    ┌──────────────┐         │       │
│  │   n8n Web    │───▶│     n8n      │────────▶│       │
│  │  Interface   │    │  (Port 5678) │         │       │
│  └──────────────┘    └──────────────┘         │       │
│                                                │       │
│  ┌────────────────────────────────────────────┘       │
│  │                                                     │
│  │  ┌──────────┐  ┌──────────┐  ┌──────────┐        │
│  └─▶│  Redis   │  │ Postgres │  │ SearxNG  │        │
│     └──────────┘  └──────────┘  └──────────┘        │
│                                                       │
└───────────────────────────────────────────────────────┘

Data Flow:

1. Cursor: Store/retrieve code context via MCP → Qdrant
2. OpenWebUI: Upload docs → RAG → Qdrant → AI answers
3. n8n: Automate workflows, integrate external APIs

---

🛠️ Advanced Usage

Custom Workflows (n8n)

1. Open http://localhost:5678
2. Login with credentials from .env file
3. Import workflow from SRC/FlowTech-AI-Complete-Workflow.json
4. Customize for your needs

Embedding Models

Default: bge-m3:567m (Ollama) - Multilingual, 1024 dimensions

Change model:

# Edit .env
RAG_EMBEDDING_MODEL=bge-large:latest  # or another model

# Restart
docker compose restart openwebui

Extend with Obsidian

Want to sync personal notes? Check the Notes Templates in Notes/_Templates/:

• vm-template.md - For virtual machines
• server-template.md - For servers
• domain-template.md - For domains

For automated sync, see companion repo: Flow-Notes-AI

---

📖 Documentation

• QUICKSTART.md - Complete setup guide
• Notes/_Templates/ - Obsidian note templates
• docs/ - Architecture & advanced topics

---

🤝 Contributing

Contributions are welcome! Please:

1. Fork the repo
2. Create a feature branch
3. Test your changes with ./init.sh
4. Submit a pull request

---

📄 License

MIT License - See LICENSE for details

---

🙏 Acknowledgments

• OpenWebUI - Amazing AI interface
• Cursor - Best AI-powered IDE
• n8n - Powerful automation platform
• Qdrant - High-performance vector database
• Ollama - Local LLM inference

---

🔗 Links

• GitHub: https://github.com/FlowTech-Lab/FlowTech-AI
• Documentation: docs/
• Issues: https://github.com/FlowTech-Lab/FlowTech-AI/issues

---

Made with ❤️ by the FlowTech-Lab community