Hermes WSL Ubuntu: deployment tool for Hermes Agent

Guide and setup scripts for running Hermes Agent with llama.cpp and Qwen3.5 on WSL2, Ubuntu and macOS

What Hermes WSL Ubuntu does

The repository documents an end-to-end local stack. Hermes Agent talks to a llama.cpp server through an OpenAI-compatible endpoint, using Qwen3.5 GGUF weights such as Qwen3.5-9B-Q5_K_M, and can also be reached through a Hermes HUDUI web interface. Configuration is done with hermes config set commands that point OPENAI_BASE_URL at http://localhost:8080/v1 and set the model name and a Telegram bot token.

A setup-wsl.sh script detects the platform and installs CUDA where applicable, Hermes, llama.cpp and the Camofox browser server, which Hermes uses for browser control. A second script, fork_ik-llamacpp.sh, installs the ik_llama.cpp fork. Separate sections cover the Hermes API server, running the server, performance and memory, troubleshooting and optimization tips, and a WSL2.md page explains the Windows side. The guide was written in March 2026 and updated in May 2026.

Key features

One-command setup script for WSL2 Ubuntu and macOS
llama.cpp build with CUDA on Linux and WSL, or Metal on macOS
Hermes configuration pointing at a local OpenAI-compatible server
Camofox browser server for Hermes browser control
Hermes HUDUI web interface and Hermes API server sections
Troubleshooting and optimization tips

When to use it

Running Hermes Agent fully on a local GPU with an open-weight model
Setting up Hermes inside WSL2 on a Windows PC
Trying the ik_llama.cpp fork as the model server

Who it is for: Users who want to run Hermes Agent against local models on a Windows (WSL2), Linux or Mac machine.

How it fits with Hermes Agent

A deployment guide built around Hermes Agent, covering its installation, configuration and a local llama.cpp backend.

How to install Hermes WSL Ubuntu

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

curl -fsSL https://raw.githubusercontent.com/metantonio/hermes-wsl-ubuntu/master/setup-wsl.sh -o setup-wsl.sh && bash setup-wsl.sh

Requirements: Linux, WSL2 Ubuntu or macOS; minimum 16 GB RAM, 4 CPU cores and a 6 GB GPU (GTX 1060 class); Node.js 18+; apt or brew; NVIDIA CUDA drivers on Linux and WSL or Apple Silicon for Metal

FAQ

What is Hermes WSL Ubuntu?

Hermes WSL Ubuntu is a guide and set of scripts for running Hermes Agent with llama.cpp and Qwen3.5 locally. It covers WSL2 Ubuntu and macOS, with GPU acceleration through CUDA or Metal.

What do I need to run Hermes WSL Ubuntu?

The README lists a minimum of 16 GB RAM, 4 CPU cores and a 6 GB GPU such as a GTX 1060, plus Node.js 18+. It recommends an RTX 30 or 40 series GPU, 12 GB or more of VRAM and 32 GB of RAM.

Is Hermes WSL Ubuntu free and open source?

The repository has no license file, so default copyright applies and you should check with the author before reuse.