Here is the complete README.md with Option B (generic placeholder approach). Copy this entire block:

# AI Workflow Stack

A containerized AI stack with n8n workflow automation, Open WebUI chat interface, LiteLLM proxy, Ollama LLM backend, and SearXNG search - all secured behind a dedicated Tailscale network with Caddy reverse proxy.

## Architecture

┌─────────────────────────────────────────────────────────┐ │ Tailscale Network (<TAILSCALE_IP>) │ │ ┌─────────────────────────────────────────────────┐ │ │ │ Caddy Reverse Proxy │ │ │ │ :4000 → LiteLLM (172.21.0.3) │ │ │ │ :8080 → Open WebUI (172.21.0.5) │ │ │ │ :5678 → n8n (172.21.0.2) │ │ │ │ :8081 → SearXNG (172.21.0.13) │ │ │ └─────────────────────────────────────────────────┘ │ │ │ │ │ ┌─────────────┐ ┌──────┴──────┐ ┌──────────────┐ │ │ │ LiteLLM │ │ Open WebUI │ │ n8n │ │ │ │ :4000 │ │ :8080 │ │ :5678 │ │ │ └──────┬──────┘ └─────────────┘ └──────┬───────┘ │ │ │ │ │ │ ┌──────┴──────┐ ┌──────────────┐ ┌────┴──────┐ │ │ │ Ollama │ │ SearXNG │ │ Sandbox │ │ │ │ :11434 │ │ :8080 │ │ Stack │ │ │ └─────────────┘ └──────────────┘ └───────────┘ │ └─────────────────────────────────────────────────────────┘

## Quick Start

### Prerequisites

- Docker & Docker Compose installed
- Tailscale account and auth key
- At least 16GB RAM (64GB+ recommended for larger models)

### 1. Clone and Configure

```bash
git clone https://github.com/KyleGolfer/ai-workflow.git
cd ai-workflow

# Copy environment template
cp .env.example .env

# Edit with your values
nano .env

2. Required Environment Variables

Variable	Description	Example
TAILSCALE_AUTH_KEY	Tailscale auth key	tskey-auth-...
LITELLM_DB_PASSWORD	PostgreSQL password	secure-password
LITELLM_MASTER_KEY	LiteLLM API key	sk-...
VENICE_API_KEY	Venice.ai API key	venice-...
N8N_USER / N8N_PASSWORD	n8n credentials	admin / password
N8N_ENCRYPTION_KEY	n8n encryption	random-string

3. Start the Stack

docker-compose up -d

4. Get Your Tailscale IP

docker exec tailscale-ai tailscale ip -4
# Returns: 100.x.x.x (your unique IP)

5. Access Services

All services available at http://<TAILSCALE_IP>:<PORT>:

Service	Port	Purpose
Open WebUI	8080	Chat interface with models
LiteLLM	4000	API key management & model routing
n8n	5678	Workflow automation
SearXNG	8081	Private search engine

Example: If your Tailscale IP is 100.70.205.60:

• Open WebUI: http://100.70.205.60:8080
• LiteLLM: http://100.70.205.60:4000

Services Overview

Open WebUI

• Web-based chat interface
• Supports multiple model providers via LiteLLM
• Document upload and RAG capabilities
• User management and permissions

LiteLLM

• Unified API for multiple LLM providers
• Cost tracking and rate limiting
• API key management
• Supports Venice.ai, OpenAI, Anthropic, etc.

n8n

• Workflow automation
• AI agent capabilities with sandboxed code execution
• Integration with 400+ services
• Visual workflow builder

Ollama

• Local LLM inference
• Runs models like Llama, Mistral, etc.
• GPU acceleration support
• Model management via CLI

SearXNG

• Privacy-focused metasearch engine
• No tracking or profiling
• Customizable search sources

Maintenance

Update Images

docker-compose pull
docker-compose up -d

View Logs

# All services
docker-compose logs -f

# Specific service
docker-compose logs -f open-webui

Backup Data

# Backup volumes
docker run --rm -v ai-workflow_open-webui-data:/data -v $(pwd):/backup alpine tar czf /backup/open-webui-backup.tar.gz -C /data .

Network Configuration

• Tailscale IP: Unique per deployment (get with docker exec tailscale-ai tailscale ip -4)
• Docker Network: 172.21.0.0/16
• Caddy proxies all traffic through Tailscale container
• No ports exposed directly to host - all traffic routed through Tailscale

Troubleshooting

Services not accessible

# Check Tailscale status
docker exec tailscale-ai tailscale status

# Verify Caddy is listening
docker exec tailscale-ai netstat -tlnp | grep caddy

# Get your Tailscale IP
docker exec tailscale-ai tailscale ip -4

Reset Tailscale

docker-compose down
sudo rm -rf /mnt/appdata/tailscale-ai
docker-compose up -d

Security Notes

• All services protected by Tailscale network (no public exposure)
• .env file contains secrets - never commit to Git
• Sandbox environment isolates untrusted code execution
• Internal Docker network prevents direct container access

License

MIT - See LICENSE file

---

## Step 2: Commit the README

**Complete this step before moving on:**

```bash
cd /mnt/appdata/ai-workflow

# Create the README file
nano README.md
# (Paste the content above, then Ctrl+X, Y, Enter)

# Stage and commit
git add README.md
git commit -m "Add comprehensive README with setup instructions

- Documented architecture with Caddy + Tailscale networking
- Step-by-step setup instructions
- Generic Tailscale IP placeholder for portability
- Service descriptions and troubleshooting guide"

# Push to GitHub
git push origin main

Confirm completion before we move to Step 1 (tidying other stacks).