███╗ ██╗ ██████╗ ██╗██████╗ ██████╗ ██████╗ ██████╗██╗ ██╗███████╗██████╗
████╗ ██║██╔═══██╗███║██╔══██╗ ██╔══██╗██╔═══██╗██╔════╝██║ ██╔╝██╔════╝██╔══██╗
██╔██╗ ██║██║ ██║╚██║██║ ██║ ██║ ██║██║ ██║██║ █████╔╝ █████╗ ██████╔╝
██║╚██╗██║██║ ██║ ██║██║ ██║ ██║ ██║██║ ██║██║ ██╔═██╗ ██╔══╝ ██╔══██╗
██║ ╚████║╚██████╔╝ ██║██████╔╝ ██████╔╝╚██████╔╝╚██████╗██║ ██╗███████╗██║ ██║
╚═╝ ╚═══╝ ╚═════╝ ╚═╝╚═════╝ ╚═════╝ ╚═════╝ ╚═════╝╚═╝ ╚═╝╚══════╝╚═╝ ╚═╝
[ PENTEST & DEV CONTAINERS | bad-antics ]
v2.0 — AI · Music · Art
Pre-configured Docker containers for security research, AI/ML, image generation, music creation, and development.
- ollama (port 11434) — Local LLM engine, runs all text models
- webui / Open WebUI (port 3080) — ChatGPT-style web interface for all models
- agent-zero (port 3100) — Autonomous AI agent with tool use + code execution
- comfyui (port 8188) — Image/logo/art generation (Stable Diffusion, SDXL, Flux)
- musicgen (port 7860) — AI music generation (Meta AudioCraft)
- pentest — Kali Linux with nmap, metasploit, burp, sqlmap
- ctf — CTF tools with pwntools, gdb, radare2, ghidra
- proxy (port 8080/8081) — mitmproxy traffic interception
- vpn (port 51820/udp) — WireGuard VPN gateway
- dev (port 3000/8000) — Python, Node, Go, Rust dev environment
- julia — Julia data science + security research
After starting Ollama, run the model pull script to download everything:
docker exec n01d-ollama bash /scripts/pull-models.sh- dolphin-mistral:7b (4.1 GB) — Eric Hartford's Dolphin, no alignment filters
- dolphin-llama3:8b (4.7 GB) — Dolphin Llama 3, fully unlocked
- dolphin-mixtral:8x7b (26 GB) — Most capable uncensored MoE model
- wizard-vicuna-uncensored:13b (7.4 GB) — Classic uncensored model
- llama2-uncensored:7b (3.8 GB) — No guardrails Llama 2
- nous-hermes2:10.7b (6.1 GB) — Powerful uncensored reasoning
- codellama:13b (7.4 GB) — Exploit dev, shellcode, reverse engineering
- deepseek-coder-v2:16b (8.9 GB) — Advanced code/vuln analysis
- phind-codellama:34b (19 GB) — Complex exploit generation
- deepseek-r1:8b (4.9 GB) — Chain-of-thought reasoning
- deepseek-r1:14b (9.0 GB) — Deep reasoning
- qwen2.5:7b (4.7 GB) — Strong multilingual reasoning
- command-r:35b (20 GB) — Advanced RAG + reasoning
- codellama:7b (3.8 GB) — Fast code generation
- starcoder2:7b (4.0 GB) — Multi-language code
- codegemma:7b (5.0 GB) — Google's code model
- qwen2.5-coder:7b (4.7 GB) — Qwen code specialist
- nomic-embed-text (274 MB) — Embeddings for Agent Zero + RAG
- all-minilm (45 MB) — Fast semantic search
cd n01d-docker
# Copy environment config
cp .env.example .env
# Build all containers
docker compose build
# Start everything (or pick what you need)
docker compose up -d
# Pull all AI models (takes a while on first run)
docker exec n01d-ollama bash /scripts/pull-models.sh
# Open your browser
# Open WebUI -> http://localhost:3080
# ComfyUI -> http://localhost:8188
# MusicGen -> http://localhost:7860
# Agent Zero -> http://localhost:3100# AI only
docker compose up -d n01d-ollama n01d-webui n01d-agent-zero
# Creative only
docker compose up -d n01d-comfyui n01d-musicgen
# Pentest only
docker compose up -d n01d-pentest n01d-proxy n01d-vpn
# Shell into pentest container
docker exec -it n01d-pentest /bin/bashEvery service binds to 0.0.0.0 so it is accessible on your LAN.
ipconfig | Select-String "IPv4"
# Look for something like 192.168.1.100 or 10.0.0.50Replace HOST_IP with the IP you found above. From any computer, phone, or tablet on the same network:
- Open WebUI —
http://HOST_IP:3080— Chat with any AI model - Agent Zero —
http://HOST_IP:3100— Autonomous AI agent - ComfyUI —
http://HOST_IP:8188— Generate images, logos, art - MusicGen —
http://HOST_IP:7860— Generate music from text - Ollama API —
http://HOST_IP:11434— Raw LLM API endpoint - mitmproxy —
http://HOST_IP:8081— Web traffic inspection UI
# Run as Administrator — opens all N01D ports
$ports = @(3080, 3100, 7860, 8080, 8081, 8188, 11434, 51820)
foreach ($port in $ports) {
New-NetFirewallRule -DisplayName "N01D Docker - Port $port" `
-Direction Inbound -Protocol TCP -LocalPort $port `
-Action Allow -Profile Private
}
# UDP for WireGuard
New-NetFirewallRule -DisplayName "N01D Docker - WireGuard UDP" `
-Direction Inbound -Protocol UDP -LocalPort 51820 `
-Action Allow -Profile Private
Write-Host "All N01D ports opened in Windows Firewall"Any app that supports Ollama (like other Open WebUI instances, Continue.dev, etc.) can point to your server:
Ollama URL: http://HOST_IP:11434
n01d-docker/
docker-compose.yml — All services defined here
.env.example — Configuration template
scripts/
pull-models.sh — Downloads all Ollama models
containers/
pentest/ — Kali security tools
dev/ — Multi-language dev env
julia/ — Julia data science
ctf/ — CTF challenge tools
proxy/ — mitmproxy
vpn/ — WireGuard
agent-zero/ — Autonomous AI agent
musicgen/ — AI music generation
config/
mitmproxy/
wireguard/
data/ — Persistent data
ollama/ — Downloaded models
open-webui/ — Chat history
agent-zero/ — Agent workdir
comfyui/ — SD models + outputs
musicgen/ — Generated music
Copy .env.example to .env and customize ports, models, etc. Each container has its own Dockerfile in containers/. All persistent data lives in data/.
If you have an NVIDIA GPU with enough VRAM, uncomment the deploy sections in docker-compose.yml for:
- Ollama — runs LLMs on GPU (much faster inference)
- ComfyUI — runs Stable Diffusion on GPU (required for decent speed)
- MusicGen — runs AudioCraft on GPU (faster generation)
You will also need NVIDIA Container Toolkit installed: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html
# Check what is running
docker compose ps
# View logs
docker compose logs -f n01d-ollama
docker compose logs -f n01d-webui
# List downloaded models
docker exec n01d-ollama ollama list
# Quick-test a model
docker exec -it n01d-ollama ollama run dolphin-mistral:7b
# Pull a single model
docker exec n01d-ollama ollama pull deepseek-r1:14b
# Stop everything
docker compose down
# Stop + remove volumes (clean slate)
docker compose down -vGitHub: https://github.com/bad-antics NullSec: https://github.com/bad-antics/nullsec
Made by bad-antics — https://github.com/bad-antics