A powerful, self-hosted Personal Knowledge Management system that rivals Obsidian. Built with Next.js and TypeScript, MarkItUp transforms markdown editing into a comprehensive second brain for organizing, linking, and discovering your thoughts.

• 🔐 User Authentication & Authorization - Secure multi-user support with JWT sessions
• 🧭 Unified AI Settings - Centralized configuration across all AI features
• 🛡️ Enhanced Security - Improved CSP headers and Docker/Ollama compatibility
• 🤖 AI Robustness - Better JSON parsing and proxy diagnostics
• � Optional Volume Mounts - Simplified deployment for managed hosting scenarios

Breaking Change: Authentication now required by default. See docs/MIGRATION_4_0.md for migration guide or set DISABLE_AUTH=true for single-user mode.

• ✨ WYSIWYG Editor - Rich text editing with TipTap
• � Intelligent Link Suggester - AI-powered wikilink suggestions with real-time analysis
• 📚 Spaced Repetition - Scientific flashcard system with FSRS algorithm
• 🤖 Universal AI Support - OpenAI, Anthropic, Gemini, or Ollama (100% local/private)
• � Semantic Search - Browser-based ML for meaning-based discovery
• 🤝 Real-time Collaboration - Multi-user editing with WebSocket sync

See CHANGELOG.md for complete version history.

git clone https://github.com/xclusive36/MarkItUp.git
cd MarkItUp
npm install
npm run dev

Visit http://localhost:3000 and start building your knowledge base!

• Bidirectional Linking - Connect notes with [[wikilinks]] and automatic backlinks
• Knowledge Graph - Visual exploration with 3D force-directed layouts
• Advanced Search - Full-text search with operators (tag:, folder:, exact phrases)
• Semantic Search - AI-powered vector search finds related content by meaning, not just keywords:
  • 🧠 Browser-Based ML - Client-side embeddings using Transformers.js (no server required!)
  • 🔍 Similarity Discovery - Find related notes based on semantic similarity
  • 📊 Progress Tracking - Real-time indexing with visual progress indicators
  • 💾 IndexedDB Storage - Efficient persistent storage for embeddings
  • ⚡ Fast Search - Sub-100ms queries for typical collections
  • 🎯 Mode Toggle - Switch between keyword and semantic search instantly
  • 🔒 100% Private - All processing happens locally in your browser
  • ⚙️ Configurable - Batch size, auto-indexing, and more in settings
• Smart Tagging - Organize with #tags and automatic indexing
• Real-time Analytics - Track your knowledge growth

All AI features work with your choice of provider (OpenAI, Anthropic, Gemini, or Ollama):

• Intelligent Link Suggester - AI finds connection opportunities in your knowledge base:
  • 🔍 Context-aware wikilink suggestions with confidence scoring
  • 🚀 Batch orphan analysis for vault-wide optimization
  • 💡 Visual context preview shows where links will be inserted
  • ⚡ Real-time suggestions while typing (optional)
  • 🧠 Pattern learning adapts to your preferences
  • 🌉 Bridge note suggestions to connect isolated clusters
  • 📊 Knowledge base health visualization
• Context-Aware AI Chat - AI assistant that understands your knowledge graph
• Writing Assistant - Advanced content analysis and improvement suggestions
• Knowledge Discovery - AI-powered gap analysis and content recommendations
• Research Assistant - Semantic search with intelligent note creation
• Multi-Provider Support - Choose once, use everywhere: OpenAI (GPT-4, GPT-3.5), Anthropic (Claude 3.5), Google Gemini (1.5 Pro/Flash), or Ollama (local AI)

MarkItUp features the most comprehensive Ollama integration in any PKM system:

• 🌊 Real-Time Streaming - Progressive response display with Server-Sent Events
• 📊 Model Management - Download, delete, and refresh models directly from the UI
• 🔌 Connection Testing - One-click server verification with detailed status (no CORS errors!)
• ⚙️ Advanced Parameters - Fine-tune context window, GPU layers, repeat penalty, and more
• 📦 Enhanced Model Details - View size, parameter count (7B/13B/70B), and quantization levels
• 💾 Zero Cost - No API keys, no usage limits, completely free local AI
• 🔒 Complete Privacy - All data stays on your machine, no external API calls
• 🚀 Model Library - Browse and install popular models (Llama 3.2, Mistral, CodeLlama, etc.)
• 🎯 Custom Servers - Connect to remote Ollama instances with saved presets
• 📈 Performance Metrics - Track tokens/second, response times, and model efficiency
• 🛡️ Seamless Integration - Built-in CORS proxy for error-free connections (no Ollama config needed!)

Getting Started with Ollama:

1. Install Ollama: brew install ollama (macOS) or visit ollama.ai
2. Start server: ollama serve
3. Pull a model: ollama pull llama3.2 or use MarkItUp's UI
4. Select "Ollama (Local)" in AI Settings - no API key needed!

Why Choose Ollama?

• 100% private - ideal for sensitive knowledge bases
• Offline capable after model download
• No monthly costs or token limits
• Full control over AI infrastructure
• Supports 50+ open-source models

• Multi-user Editing - Real-time collaborative editing with conflict resolution
• Live Presence - See other users' cursors and selections
• WebSocket Sync - Low-latency synchronization via Socket.IO
• Session Sharing - Easy document sharing with secure links

• Rich Plugin API - Custom functionality, commands, and processors
• Settings Management - Configurable plugin settings with persistence
• Content Processors - Transform content during import/export
• Event System - Plugin communication through events

• WYSIWYG Editor - Rich text editing with TipTap, seamless markdown conversion
• Multi-mode Editor - Edit, preview, split-view, or WYSIWYG modes
• LaTeX Math - Inline $math$ and block $$math$$ equations
• TikZ Diagrams - Create vector graphics directly in markdown
• GitHub Flavored Markdown - Tables, task lists, and more

🌐 Web-Native: Access from any device with a browser - no app installation required

🏠 Complete Privacy: Self-hosted solution means your data never leaves your control

🔗 Built for Connection: Native wikilinks, backlinks, and graph visualization

⚡ Modern Stack: Built with Next.js 16 and TypeScript for speed and reliability

🆓 Truly Free: Open-source with no licensing fees or feature restrictions

🚀 Extensible: Modern plugin architecture for unlimited customization

MarkItUp offers the most flexible AI integration in any PKM system:

Choose the right provider for your needs:

• OpenAI - Most capable general-purpose AI (GPT-4)
• Anthropic - Long context windows, ethical AI (Claude 3.5)
• Gemini - Most cost-effective with massive 1M token context (Gemini 1.5 Pro/Flash)
• Ollama - Complete privacy, zero cost, offline capable (50+ models)

⚠️ REQUIRED: Generate secure secrets before deploying:

# Generate JWT_SECRET
openssl rand -base64 32

# Generate ENCRYPTION_KEY
openssl rand -hex 16

Create docker-compose.yml:

version: "3.8"

services:
  markitup:
    container_name: markitup
    ports:
      - 3000:3000
    # volumes:
      # Optional: Uncomment to persist markdown files on host filesystem
      # Note: Requires chmod 777 ./markdown for write permissions
      # For managed hosting or ephemeral deployments, leave commented
      # - ./markdown:/app/markdown
    environment:
      - PORT=3000
      - HOSTNAME=0.0.0.0
      - NODE_ENV=production
      # REQUIRED: Replace with your generated secrets
      - JWT_SECRET=your-generated-jwt-secret-here
      - ENCRYPTION_KEY=your-generated-encryption-key-here
      # Optional: Disable auth for testing (NOT FOR PRODUCTION)
      # - DISABLE_AUTH=false
    restart: unless-stopped
    image: ghcr.io/xclusive36/markitup:latest

docker compose up -d

# Generate secrets
JWT_SECRET=$(openssl rand -base64 32)
ENCRYPTION_KEY=$(openssl rand -hex 16)

# Run without persistent storage (default)
docker run --name markitup -p 3000:3000 \
  -e JWT_SECRET="$JWT_SECRET" \
  -e ENCRYPTION_KEY="$ENCRYPTION_KEY" \
  --restart unless-stopped \
  ghcr.io/xclusive36/markitup:latest

# OR with persistent storage
docker run --name markitup -p 3000:3000 \
  -v ./markdown:/app/markdown \
  -e JWT_SECRET="$JWT_SECRET" \
  -e ENCRYPTION_KEY="$ENCRYPTION_KEY" \
  --restart unless-stopped \
  ghcr.io/xclusive36/markitup:latest

By default, data lives in the container. To persist data on your host:

1. Uncomment the volume mount in docker-compose.yml or add -v ./markdown:/app/markdown to docker run
2. Set permissions:

# Easy method (all users can write)
chmod -R 777 ./markdown

# Secure method (container user only)
sudo chown -R 1001:1001 ./markdown

For managed hosting, skip volume mounts and implement alternative storage (S3, database, etc.)

MarkItUp supports both multi-user authenticated mode (default, recommended) and single-user unauthenticated mode for personal self-hosted deployments.

Features:

• ✅ Secure user isolation - each user has their own /markdown/user_{id}/ directory
• ✅ User management with signup, login, and session handling
• ✅ Per-user quotas and usage tracking
• ✅ Audit logs showing who modified what
• ✅ Collaborative features (sharing, permissions)

Requirements:

• Set JWT_SECRET and ENCRYPTION_KEY environment variables
• Users must sign up and log in

For personal self-hosted instances where you're the only user:

# docker-compose.yml
environment:
  - DISABLE_AUTH=true

What This Does:

• ✅ No signup/login required - instant access
• ✅ All files stored in /markdown/ root directory
• ✅ Simpler setup for single-user scenarios
• ⚠️ No user isolation - all data is shared
• ⚠️ No access control - anyone with network access can read/write
• ⚠️ No audit trail - can't track who changed what
• ⚠️ Limited features - collaboration and sharing disabled

⚠️ SECURITY WARNINGS:

🚨 DISABLE_AUTH=true means:
   • Anyone who can reach your server can access ALL your notes
   • No password protection whatsoever
   • No way to restrict who can create, edit, or delete files
   • All AI Chat file operations affect the shared directory
   • Should ONLY be used for:
     - Personal single-user deployments
     - Localhost-only access
     - Development/testing
     - Networks you completely trust

When to Use Each Mode:

Best Practices for DISABLE_AUTH=true:

1. Network Isolation: Only expose on localhost or trusted LAN
2. Firewall Rules: Block external access at network level
3. Reverse Proxy: Use nginx/Caddy with basic auth if internet-facing
4. Regular Backups: No audit log means you can't recover from mistakes
5. Single User Only: Never use for multi-user scenarios

Migration: You can switch between modes anytime. Existing data in /markdown/ will work in both modes, though user-specific folders (/markdown/user_*/) are only used in multi-user mode.

We welcome contributions! Please see our contribution guidelines for details.

• 🐛 Bug Reports: Open an issue
• 💡 Feature Requests: Start a discussion
• 🔧 Pull Requests: Fork, develop, and submit PRs

MIT License - see LICENSE file for details.

⭐ Star this repo if you find MarkItUp useful!

Built with ❤️ by the MarkItUp community