Experimental AI/ML and specialized scripts for Proxmox VE
Originally created by tteck, maintained by the community, extended by hereteks
⚙️ The Heretek's Forge
"The Omnissiah guides our scripts, but innovation demands we venture beyond established doctrine."
While the orthodox community-scripts/ProxmoxVE maintains the sacred canon of stable, community-vetted scripts, this Heretek repository serves as a forge for experimental and specialized tools that may one day prove worthy of the main temple.
What makes this a Heretek forge:
- 🧪 Experimental AI/ML Scripts — Cutting-edge AI tooling before mainstream adoption
- 🔧 Niche Applications — Specialized tools for specific use cases
- ⚡ Rapid Iteration — Faster release cycles for emerging technologies
- 🔄 Upstream Path — Successful scripts may ascend to the orthodox repository
🏛️ Relationship to the Orthodox Repository
This fork maintains synchronization with community-scripts/ProxmoxVE while adding experimental scripts that may not yet meet the stability requirements for upstream inclusion. See how this fork works.
| Aspect | Orthodox (Upstream) | Heretek (This Fork) |
|---|
| Stability | Production-ready | Experimental/Bleeding-edge |
| Focus | Broad community needs | AI/ML & specialized tools |
| Release Cycle | Stable & tested | Rapid iteration |
| Best For | General users | Early adopters |
Scripts that prove stable and gain community interest may be submitted upstream to join the orthodox canon.
🧪 Heretek Scripts
These scripts are exclusive to this forge, focusing on AI/ML workloads and specialized applications:
AI & Machine Learning
| Script | Description | Resources |
|---|
| Hermes | Nous Research Hermes AI Agent (with Docker & CUA support) | 4 CPU / 8GB RAM |
| MCPHub | Model Context Protocol Hub for AI tooling integration | 2 CPU / 2GB RAM |
| Strata | Inference engine for Qwen3.8-Flash-Next with a Strix Halo HIP engine, OpenAI-compatible API | 8 CPU / 64GB RAM + Strix Halo GPU |
| GSQHalo.cpp | llama.cpp fork for GSQ-RCO Qwen3.8-Flash-Next on Strix Halo, own ROCm 7.14.1 | 8 CPU / 64GB RAM + Strix Halo GPU |
| SwarmUI | Stable Diffusion Swarm interface | 8 CPU / 16GB RAM + GPU |
| LlamaCPP | High-performance LLM inference engine with GPU acceleration | 4 CPU / 8GB RAM + GPU |
| Gufo | Local inference engine tuned for AMD Strix Halo, OpenAI-compatible API | 8 CPU / 32GB RAM + Strix Halo GPU |
| Kyojin | ExLlamaV3-based engine for Strix Halo serving Yamz's EXL3 Qwen3.8, GLM-5.3 and MiMo packs, OpenAI-compatible API | 8 CPU / 120GB RAM + Strix Halo GPU |
| OpenClaw | AI-powered assistant and automation platform |
Media & Gaming
| Script | Description | Resources |
|---|
| Agregarr | Media aggregation and streaming | 2 CPU / 4GB RAM |
| Drop | Gaming and media server | 2 CPU / 4GB RAM |
🖥️ Virtual Machine Scripts
Complete VM deployment scripts for various operating systems and specialized appliances:
Operating Systems
Network & Security
Storage & Cloud
Home Automation & Docker
⚠️ Note: Heretek scripts may require more resources or additional configuration. Check individual script documentation for requirements.
🎯 Key Features
📋 Requirements
📥 Getting Started
Choose your preferred installation method:
Method 1: One-Click Web Installer
The fastest way to get started:
- Visit helpful-scripts.com 🌐
- Search for your desired script (e.g., "MCPHub", "SwarmUI", "LlamaCPP")
- Copy the bash command displayed on the script page
- Open your Proxmox Shell and paste the command
- Press Enter and follow the interactive prompts
Method 2: Direct Script Execution
For Heretek scripts, use the Heretek-AI repository:
# Example: Install MCPHub
bash -c "$(curl -fsSL https://raw.githubusercontent.com/Heretek-AI/ProxmoxVE/main/ct/mcphub.sh)"
Method 3: PVEScripts-Local
Install a convenient script manager directly in your Proxmox UI:
bash -c "$(curl -fsSL https://raw.githubusercontent.com/Heretek-AI/ProxmoxVE/main/ct/pve-scripts-local.sh)"
This adds a menu to your Proxmox interface for easy script access without visiting the website.
📖 Learn more: ProxmoxVE-Local Repository
💬 Join the Community
🛠️ Contribute
👉 Check our Contributing Guidelines to get started
❤️ Support the Project
This project is maintained by volunteers in memory of tteck. Your support helps us maintain infrastructure, improve documentation, and give back to important causes.
🎗️ 30% of all donations go directly to cancer research and hospice care
📈 Project Statistics
📜 License
This project is licensed under the MIT License — feel free to use, modify, and distribute.
