My AI-Powered Workflow: Integration of MCP, Docker and N8N for Obsidian

Objective

Automate the organization and processing of Obsidian notes using AI accessible via API, orchestrated with Docker, MCP (Modular Copilot Platform), and N8N workflows.

Why This Workflow?

The Problem

Manual Note Management Challenges: - Notes scattered across multiple topics - Inconsistent language (French/English mix) - No automated categorization - Time-consuming manual organization - Difficult to find related content

The Solution

AI-Powered Automation: - Automatic note organization by theme - Language detection and translation - Intelligent content grouping - Automated index generation - Scheduled processing workflows

Prerequisites

Software Requirements

Essential Tools: - Docker installed and running - MCP (Modular Copilot Platform) deployed - N8N installed (via Docker or standalone) - Obsidian vault accessible locally

Optional Tools: - LLM model (Mistral, Llama, etc.) - Ollama for local model hosting - Git for version control

Architecture Overview

Obsidian Vault (Markdown Files)
    ↓
N8N Workflow (File Reader)
    ↓
MCP API (AI Processing)
├── Organize by theme
├── Translate content
├── Generate summaries
└── Create indexes
    ↓
N8N Workflow (File Writer)
    ↓
Obsidian Vault (Organized Files)

Step 1: Deploy MCP with Docker

Quick Start

# Pull and run MCP container
docker run -d \
  --name mcp \
  -p 5000:5000 \
  -v $(pwd)/mcp_data:/app/data \
  ghcr.io/modularcopilot/mcp:latest

# Verify MCP is running
curl http://localhost:5000/health

# Expected response:
# {"status": "healthy", "version": "1.0.0"}

Docker Compose (Recommended)

# docker-compose.yml
version: '3.8'

services:
  mcp:
    image: ghcr.io/modularcopilot/mcp:latest
    container_name: mcp-server
    restart: unless-stopped
    ports:
      - "5000:5000"
    volumes:
      - ./mcp_data:/app/data
      - ./models:/app/models
    environment:
      - MCP_MODEL=mistral:7b
      - MCP_API_PORT=5000
      - OLLAMA_HOST=http://host.docker.internal:11434
    networks:
      - homelab

  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    restart: unless-stopped
    ports:
      - "11434:11434"
    volumes:
      - ollama_data:/root/.ollama
    networks:
      - homelab

networks:
  homelab:
    driver: bridge

volumes:
  ollama_data:

# Deploy services
docker-compose up -d

# Check logs
docker-compose logs -f mcp

# Pull model in Ollama
docker exec ollama ollama pull mistral:7b

Verify MCP API

# Test API endpoint
curl -X POST http://localhost:5000/api/prompt \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Summarize this text: Proxmox is a virtualization platform.",
    "model": "mistral:7b"
  }'

Step 2: Prepare AI Prompts

Example Prompt Templates

Organize Notes by Language:

{
  "prompt": "Unify all my Obsidian notes into English. Translate French titles, group similar content, and create a thematic index."
}

Categorize Notes:

{
  "prompt": "Analyze these Obsidian notes and categorize them into: Infrastructure, Security, Automation, and Reference. Return structured JSON."
}

Generate Summary:

{
  "prompt": "Create a concise summary (max 100 words) of this technical article while preserving key commands and configurations."
}

Extract Tags:

{
  "prompt": "Extract relevant tags from this note. Return comma-separated tags suitable for Obsidian YAML frontmatter."
}

Step 3: Create N8N Workflow

Workflow Components

Complete Workflow Structure:

1. Schedule Trigger (Daily at 2 AM)
    ↓
2. Read Obsidian Files
    ↓
3. Filter (Only .md files)
    ↓
4. Extract YAML Frontmatter
    ↓
5. HTTP Request to MCP
    ↓
6. Process AI Response
    ↓
7. Update/Create Organized Files
    ↓
8. Send Notification (Telegram/Email)

Workflow Configuration

Node 1: Schedule Trigger

{
  "node": "Schedule Trigger",
  "type": "n8n-nodes-base.cron",
  "parameters": {
    "rule": {
      "interval": [
        {
          "field": "cronExpression",
          "expression": "0 2 * * *"
        }
      ]
    }
  }
}

Node 2: Read Files

{
  "node": "Read Files",
  "type": "n8n-nodes-base.readBinaryFiles",
  "parameters": {
    "filePath": "/path/to/obsidian/vault",
    "filePattern": "*.md"
  }
}

Node 3: HTTP Request to MCP

{
  "node": "MCP API Call",
  "type": "n8n-nodes-base.httpRequest",
  "parameters": {
    "method": "POST",
    "url": "http://localhost:5000/api/prompt",
    "authentication": "none",
    "sendHeaders": true,
    "headerParameters": {
      "parameters": [
        {
          "name": "Content-Type",
          "value": "application/json"
        }
      ]
    },
    "sendBody": true,
    "bodyParameters": {
      "parameters": [
        {
          "name": "prompt",
          "value": "Organize my notes Obsidian by theme and language."
        },
        {
          "name": "notes",
          "value": "={{ $json[\"content\"] }}"
        }
      ]
    }
  }
}

Node 4: Write File

{
  "node": "Save Organized Notes",
  "type": "n8n-nodes-base.writeFile",
  "parameters": {
    "fileName": "/vault/Index/organization.md",
    "data": "={{ $json[\"response\"] }}"
  }
}

Step 4: HTTP Request Configuration

Detailed Request Setup

Method: POST

URL: http://localhost:5000/api/prompt

Headers:

{
  "Content-Type": "application/json",
  "Authorization": "Bearer YOUR_MCP_API_KEY"
}

Body:

{
  "prompt": "Organize mes notes Obsidian par thème et langue.",
  "notes": "{{ $json[\"content\"] }}",
  "options": {
    "temperature": 0.7,
    "max_tokens": 2048,
    "output_format": "markdown"
  }
}

Response Processing

Expected Response:

{
  "status": "success",
  "organized_content": "# Infrastructure\n\n## Proxmox\n- Note 1\n- Note 2\n\n# Security\n...",
  "metadata": {
    "notes_processed": 45,
    "categories_created": 6,
    "languages_detected": ["en", "fr"]
  }
}

Step 5: Save Results to Obsidian

File Organization Strategy

Directory Structure:

Vault/
├── Index/
│   ├── organization.md (auto-generated)
│   ├── by-theme.md
│   └── by-language.md
├── Infrastructure/
│   ├── EN/
│   └── FR/
├── Security/
│   ├── EN/
│   └── FR/
└── Automation/
    ├── EN/
    └── FR/

Index File Template

# 📚 Automated Organization Index
*Generated: {{date}}*

## By Theme
[[Infrastructure]] | [[Security]] | [[Automation]] | [[Reference]]

## By Language
🇬🇧 [[English Notes]] | 🇫🇷 [[French Notes]]

## Statistics
- Total notes: {{total_count}}
- Categories: {{category_count}}
- Last updated: {{timestamp}}

---
*Automated by MCP + N8N workflow*

Tips and Best Practices

Use Language Tags

In Obsidian Notes:

---
tags: [homelab, #en]
language: en
---

# My Infrastructure Note
Content here...

Create Multilingual Index

Index Structure:

# 🌍 Multilingual Index

## Infrastructure
- 🇬🇧 [[Proxmox Setup EN]]
- 🇫🇷 [[Installation Proxmox FR]]

## Security
- 🇬🇧 [[pfSense Configuration EN]]
- 🇫🇷 [[Configuration pfSense FR]]

Enhance Prompts

Advanced Prompt Examples:

{
  "prompt": "Organize notes by theme and language. For each note:\n1. Detect language\n2. Assign category\n3. Generate 3-5 tags\n4. Create 1-sentence summary\n5. Link related notes\nReturn structured markdown."
}

Advanced Workflow Features

Conditional Processing

Process Only Modified Files:

// In N8N Function node
const lastRun = $node["Get Last Run"].json.timestamp;
const fileModified = new Date($json.modified);

if (fileModified > lastRun) {
  return $json; // Process this file
}

return null; // Skip this file

Error Handling

Retry Logic:

{
  "node": "Error Handler",
  "onError": "continueErrorOutput",
  "retryOnFail": true,
  "maxTries": 3,
  "waitBetweenTries": 1000
}

Notification

Telegram Notification:

{
  "node": "Send Summary",
  "type": "n8n-nodes-base.telegram",
  "parameters": {
    "chatId": "YOUR_CHAT_ID",
    "text": "✅ Notes organized!\n\nProcessed: {{notes_count}}\nCategories: {{categories}}\nTime: {{duration}}s"
  }
}

Resources

Documentation

Example Workflows

GitHub Repository:

git clone https://github.com/username/obsidian-ai-workflows
cd obsidian-ai-workflows/n8n

Import into N8N:

Settings → Import from File
→ Select workflow.json
→ Activate workflow

Troubleshooting

MCP Connection Issues

# Check MCP is running
docker ps | grep mcp

# View logs
docker logs -f mcp-server

# Test API
curl http://localhost:5000/health

N8N Can’t Reach MCP

# If MCP is in Docker, use docker network
docker network inspect homelab

# Or use host network mode
docker run --network host mcp

File Permission Errors

# Fix Obsidian vault permissions
chmod -R 755 /path/to/vault

# For Docker volume access
chown -R 1000:1000 /path/to/vault

Conclusion

This AI-powered workflow transforms Obsidian note management from manual drudgery to automated intelligence. By combining MCP’s AI capabilities with N8N’s workflow orchestration, you create a self-organizing knowledge base that continuously improves.

Key Benefits: - Automated note organization - Multilingual content management - Intelligent categorization - Time savings (hours → minutes) - Consistent structure

Next Steps: - Fine-tune prompts for your note style - Add custom categories - Integrate with other tools (Notion, Evernote) - Create visualization dashboards

Automate your knowledge, amplify your productivity.

Ai Powered Workflow Mcp Docker N8n Obsidian