Hermes

Hermes is Nous Research's terminal-first AI agent CLI. Its user configuration lives under ~/.hermes/: runtime settings in config.yaml and secrets in .env.

Install

macOS / Linux

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

Windows PowerShell

iex (irm https://hermes-agent.nousresearch.com/install.ps1)

Open a new terminal after installation if hermes is not immediately on your PATH.

Configure with the CLI

Create a named key in the NexoAI dashboard, then register nexoai as a custom OpenAI-compatible Hermes provider. These commands keep the secret in ~/.hermes/.env and reference it from config.yaml:

hermes config set providers.nexoai.name NexoAI
hermes config set providers.nexoai.base_url https://api.nexoai.ma/v1
hermes config set providers.nexoai.key_env NEXOAI_API_KEY
hermes config set providers.nexoai.transport openai_chat
hermes config set model.provider nexoai
hermes config set model.default gpt-5.6-sol
hermes config set NEXOAI_API_KEY sk-your-nexoai-key

providers.nexoai.base_url is the versioned OpenAI-compatible API base. NEXOAI_API_KEY is routed to Hermes's .env because it is a credential variable.

Match key access to the model

Use a named key with access to the model family you select. For the main agent, gpt-5.6-sol, claude-opus-4-8, and qwen3.7-plus are current NexoAI catalog options with strong agent and coding capability.

Configure manually

You can make the same change by merging this into ~/.hermes/config.yaml:

model:
  default: gpt-5.6-sol
  provider: nexoai

providers:
  nexoai:
    name: NexoAI
    base_url: https://api.nexoai.ma/v1
    key_env: NEXOAI_API_KEY
    transport: openai_chat

Then add the named NexoAI key to ~/.hermes/.env:

NEXOAI_API_KEY=sk-your-nexoai-key

On macOS or Linux, restrict the secret file to your user:

chmod 600 ~/.hermes/.env

Keep .env private

Do not put a live key directly in a shared config.yaml or dotfiles repository. Revoke the named key in the NexoAI dashboard if it is ever exposed.

Verify

Start Hermes:

hermes

Inside the terminal UI, use /model to confirm the nexoai provider and select a configured NexoAI model. Send a small prompt, then check that usage appears under the same named key in the dashboard.

Handy commands

Command	Use
hermes	Start the interactive terminal agent
hermes model	Choose or change the provider and model
hermes dashboard	Open the local Hermes dashboard
hermes config set <key> <value>	Set one configuration or credential value
hermes doctor	Diagnose installation, provider, and runtime issues
hermes update	Update Hermes to the latest release

If model discovery does not list the expected ID, enter the exact ID from the NexoAI catalog and run hermes doctor before changing the gateway base.

Next: configure OpenClaw ->