---

🌐 Project Mycelium

Local-first knowledge graph with offline AI, CRDT sync, and encrypted storage. Zero cloud dependency.

Like fungal mycelium, ideas propagate underground — resilient, private, and interconnected.

---

✦ Features

• Local-first & encrypted storage – AES-256-GCM, SQLite, content-addressed SHA-256
• CRDT sync – Yjs for conflict-free collaboration
• Offline AI – DeepSeek R1 Distill (Qwen-7B) integrated locally, no cloud dependency
• Interactive visualizations – D3.js graphs
• Full offline app – no demo placeholder; all functionality is production-ready

---

✦ Quick Start

git clone https://github.com/Zorvia/project-mycelium.git
cd project-mycelium
npm run dev

• Frontend → http://localhost:3000
• Backend → http://localhost:8000
• API docs → http://localhost:8000/docs

Docker:

docker build -t mycelium:app .
docker run -p 8000:8000 mycelium:app

---

✦ DeepSeek AI Integration

• Model files (model-00001-of-000002.safetensors + model-00002-of-000002.safetensors) reside in src/backend/deepseek_adapter/DeepSeek-R1-Distill-Qwen-7B/
• Initialize the AI adapter in Python:

from mycelium.ai.deepseek_adapter import initialize

adapter = initialize("src/backend/deepseek_adapter/DeepSeek-R1-Distill-Qwen-7B")
print(adapter.status())
response = adapter.predict("Explain local-first storage in simple terms")
print(response)

• Supports streaming token generation, status checks, and safe shutdown.
• Fully offline, memory-efficient, atomic extraction for large .safetensors files.

---

✦ Architecture

graph LR
    UI["React UI / Graph Explorer"]
    AI["DeepSeek R1 Distill AI"]
    CRDT["Yjs CRDT"]
    API["FastAPI Backend"]
    DB["SQLite DB"]
    CHUNK["Chunking + AES-256-GCM"]
    P1["Peer A"]
    P2["Peer B"]

    UI --> API
    UI --> AI
    UI --> CRDT
    API --> CHUNK --> DB
    CRDT <-->|WebRTC| P1
    CRDT <-->|WebRTC| P2

Loading

---

✦ Project Structure

project-mycelium/
├── src/backend/
│   └── deepseek_adapter/      # DeepSeek R1 adapter module
├── src/frontend/
├── scripts/
├── docs/
├── tests/
├── Dockerfile
├── docker-compose.yml
├── package.json
├── requirements.txt
└── LICENSE.md

---

✦ Philosophy

• Your data is yours — fully encrypted, no cloud
• Knowledge should be free — open-source and educational
• Privacy by default — encryption is standard
• Offline AI for local intelligence — no external calls

---

✦ Contributing

• Maintain clear, modular code
• Follow project style & architecture
• Open to all contributions via Pull Requests (fork → feature branch → PR)

Repository: GitHub

---

✦ License

Zorvia Public License v2.0

---