SanctumWriter ✍️

Your private sanctuary for writing with AI — 100% Local, 100% Yours

A local-first markdown editor that uses your own LLMs (Ollama/LM Studio) as a collaborative writing companion. Like Cursor for code, but for prose.

---

🔐 Why SanctumWriter?

Feature	SanctumWriter	Other AI Writers
Privacy	✅ 100% local - nothing leaves your machine	❌ Data sent to cloud servers
Cost	✅ Free forever (uses your local LLMs)	❌ Monthly subscriptions
Internet	✅ Works offline	❌ Requires connection
Your Data	✅ Stored locally, you control it	❌ Stored on company servers
Open Source	✅ MIT License	❌ Usually closed source

---

🆚 SanctumWriter vs SanctumWriter Pro

Feature	SanctumWriter	SanctumWriter Pro
LLM Providers	Ollama, LM Studio (local only)	Local + OpenRouter, OpenAI, Anthropic, Google, xAI
Privacy	100% Local	Choice of local or cloud
Cost	Free	Free + API costs
Best For	Privacy-focused writers	Writers wanting access to frontier models
Port	localhost:3125	localhost:3130

💡 Choose SanctumWriter if privacy is paramount and you're happy with local models.
💡 Choose Pro if you need GPT-4, Claude, or other cloud models.

---

✨ Features

Core Writing

• 📝 Rich Markdown Editor — Full-featured editor with syntax highlighting (CodeMirror 6)
• 👁️ Live Preview — See rendered markdown as you type
• 📁 Workspace Browser — Navigate and manage your documents (Obsidian-compatible!)
• 💾 Auto-Save — Never lose your work

AI Writing Companion

• 🤖 Agentic Editing — AI directly modifies your document (no copy/paste!)
• 🎯 Selection-Aware — Highlight text and ask the AI to rewrite just that section
• 💬 Contextual Chat — AI sees your full document and selection
• 🔧 Hardware Optimization — Auto-detects your GPU and optimizes settings

Advanced Features

• 👥 Council of Writers — Multiple AI reviewers analyze your work
• 🔍 Research Integration — Search with SearXNG for fact-checking
• 📊 Quality Assurance — Hallucination detection, fact verification, AI artifact removal
• 📋 Writing Workflow — Guided checklist from outline to polish
• 📈 Readability Metrics — Flesch-Kincaid and other scores
• 🎯 Focus Mode — Distraction-free writing
• 📚 Citations & Bibliography — Key-based citation management
• 📤 Export — PDF, DOCX, HTML, TXT formats
• 🧠 RAG Knowledge Base — Use your documents as AI context
• 💭 Session Memory — AI remembers your writing preferences
• 🎨 Image Studio — Generate images via local ComfyUI (Stable Diffusion)

---

🚀 Quick Start

Prerequisites & Dependencies

SanctumWriter integrates with several local services. Here's what you need:

System Requirements

Requirement	Version	Install Link
Node.js	18+	nodejs.org
npm	Included with Node.js	—

LLM Backend (Pick One)

#	Service	Purpose	Install Link
1	Ollama ⭐	Local LLM inference (recommended)	ollama.ai
—	OR LM Studio	Alternative local LLM with GUI	lmstudio.ai

Optional (For Advanced Features)

#	Service	Purpose	Install Link	Requires
2	Docker Desktop	Container runtime for search services	docker.com	—
3	SearXNG	Privacy-focused web search	GitHub	Docker
4	Perplexica	AI-powered search with summaries	GitHub	Docker + Ollama
5	ComfyUI	Local image generation (Stable Diffusion)	GitHub	GPU recommended

Installation Order

REQUIRED (Native Install):
─────────────────────────
1. Node.js (for SanctumWriter)
2. Ollama OR LM Studio (for AI) ← Install directly, NOT in Docker

OPTIONAL (Docker-based):
────────────────────────
3. Docker Desktop (only if using search features below)
4. SearXNG (privacy search - runs in Docker)
5. Perplexica (AI search - runs in Docker)

FINALLY:
────────
6. SanctumWriter

⚠️ Note: Install Ollama/LM Studio natively on your machine (not in Docker). While Docker versions exist, native installation gives better performance and GPU access.

---

Setting Up Ollama (Recommended)

# 1. Install Ollama from https://ollama.ai

# 2. Pull a writing-focused model
ollama pull qwen3:latest

# 3. Start the server (runs on port 11434)
ollama serve

Setting Up LM Studio (Alternative)

1. Download from lmstudio.ai
2. Load a model (e.g., Llama 3, Mistral, Qwen)
3. Go to Local Server tab → Start Server (runs on port 1234)

Setting Up Search Services (Optional)

SearXNG (Privacy-focused search):

# Using Docker
docker run -d --name searxng -p 4000:8080 searxng/searxng

Perplexica (AI-powered search):

# Clone and follow setup instructions
git clone https://github.com/ItzCrazyKns/Perplexica.git
cd Perplexica
# See their README for Docker setup (runs on port 3000)

---

Installing SanctumWriter

Option A: Standard Installation (npm)

# Clone the repo
git clone https://github.com/lafintiger/SanctumWriter.git
cd SanctumWriter

# Install dependencies
npm install

# Start the app
npm run dev

Open http://localhost:3125 in your browser.

Option B: Docker Installation 🐳

Quick Start (uses Ollama on your host machine):

# Clone the repo
git clone https://github.com/lafintiger/SanctumWriter.git
cd SanctumWriter

# Build and run
docker-compose up -d

# View logs
docker-compose logs -f

With Ollama in Docker (no local Ollama needed):

# Start app + Ollama container
docker-compose --profile ollama up -d

# Pull a model into the container
docker exec sanctum-ollama ollama pull qwen3:latest

Development with hot-reloading:

docker-compose -f docker-compose.dev.yml up

Docker Command	Description
docker-compose up -d	Start app (connects to host Ollama)
docker-compose --profile ollama up -d	Start app + Ollama container
docker-compose down	Stop all services
docker-compose logs -f	View live logs
docker-compose build --no-cache	Rebuild after code changes

Environment Variables:

# Create .env file for custom settings
OLLAMA_URL=http://host.docker.internal:11434
LMSTUDIO_URL=http://host.docker.internal:1234
DEFAULT_PROVIDER=ollama
DEFAULT_MODEL=llama3

💡 Tip: Your documents are persisted in ./documents folder and LanceDB data in a Docker volume.

---

📖 Usage

Basic Editing

1. Click + to create a new document
2. Write markdown in the editor
3. Documents auto-save as you type

AI Assistance

1. Type a message in the chat panel
2. The AI sees your document and any selected text
3. Ask for help: "Make this more engaging" or "Expand this section"

Selection-Based Editing

1. Highlight text in the editor
2. Chat shows "Selection active"
3. Ask: "Rewrite this" or "Make it more concise"
4. AI directly modifies just the selected text

Council of Writers (Multi-Model Review)

1. Open Settings → Council Configuration
2. Enable reviewers (Style, Clarity, Fact-checker, etc.)
3. Click Start Council Review
4. Review suggestions in the Review Document

---

⚙️ Configuration

Service URLs (Settings Panel)

Configure your local services in the Settings modal. All URLs are customizable if you use non-default ports:

Service	Default URL	Purpose	Project
Ollama	http://localhost:11434	Local LLM inference	ollama.ai
LM Studio	http://localhost:1234	Alternative local LLM	lmstudio.ai
SearXNG	http://localhost:4000	Privacy-focused search	GitHub
Perplexica	http://localhost:3000	AI-powered search	GitHub
ComfyUI	http://localhost:8188	Image generation	GitHub

💡 Tip: If running services in Docker, use localhost:PORT — SanctumWriter runs on your host machine and can reach Docker containers via localhost.

Workspace Folder

Set your working directory in Settings → Workspace. Works great with Obsidian vaults!

---

⌨️ Keyboard Shortcuts

Shortcut	Action
Ctrl/Cmd + S	Save document
Ctrl/Cmd + Z	Undo
Ctrl/Cmd + Shift + Z	Redo
Ctrl/Cmd + F	Find in document
Escape	Toggle Focus Mode

---

🛠️ Tech Stack

• Framework: Next.js 14 (App Router)
• Editor: CodeMirror 6
• Styling: Tailwind CSS
• State: Zustand
• Vector DB: LanceDB (for RAG)
• LLM: Ollama / LM Studio

---

🐛 Troubleshooting

"Ollama not available"

# Make sure Ollama is running
ollama serve

# Check it's accessible
curl http://localhost:11434/api/tags

Models not showing

# Pull a model first
ollama pull qwen3:latest

# Or for a smaller model
ollama pull gemma3:4b

Port conflict

The app runs on port 3125 by default. If you need a different port, modify package.json.

---

📄 License

MIT - See LICENSE for details.

---

🙏 Acknowledgments

Built with ❤️ for writers who value their privacy.

---

SanctumWriter — Your words. Your sanctuary. Your privacy.

Report Bug · Request Feature