Skip to main content

Hermes Agent

4i.codes8/13/26Less than 1 minute

Who this guide is for

This guide is for users who want to run hermes directly on their own computer. Windows users must run Hermes inside WSL.

1. Install Hermes

Run the installer in macOS, Linux, or a WSL terminal:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Hermes is normally installed under ~/.hermes. Restart the terminal if the hermes command is not immediately available.

2. Connect Hermes to 4i.codes

Open ~/.hermes/config.yaml and configure the model:

model:
  provider: custom
  default: gpt-5.6-sol
  base_url: https://api.4i.codes
  api_mode: chat_completions

Then add your key to ~/.hermes/.env:

OPENAI_API_KEY=YOUR_4ICODES_KEY

Create a key first if needed: API Keys and Access.

3. Apply the cache compatibility patch

The patch currently targets the GPT Codex service group.

git clone https://github.com/foryourhealth111-pixel/hermes-codex-proxy-cache-compat.git
cd hermes-codex-proxy-cache-compat
bash scripts/apply_patches.sh ~/.hermes/hermes-agent
bash scripts/install_skill.sh ~/.hermes

This aligns Hermes requests with Codex caching behavior and can improve cache hit rates.

4. Test the installation

hermes

Ask a short question. If Hermes returns a response, the basic setup is complete. For key, address, or model errors, recheck config.yaml, .env, and the key's service group.

After Hermes updates, you may need to apply the compatibility patch again.

4i.codes