Skip to main content
Hermes Agent is Nous Research’s terminal AI assistant with a unified TUI interface. Connect it to ClearMaas via its YAML config file or environment variables to access OpenAI and Anthropic models through a single endpoint.

​
Prerequisites

  • Hermes Agent installed (pip install hermes-agent or via installer)
  • A ClearMaas API key starting with sk-

​
Config file locations

All settings live in ~/.hermes/:
FilePurpose
config.yamlModel, terminal, agent, and provider settings
.envAPI keys and secrets
auth.jsonOAuth provider credentials

​
Setup via config.yaml

Add to ~/.hermes/config.yaml: 
model:
  default: "anthropic/claude-sonnet-4.6"
  provider: "custom"
  base_url: "https://api.clearmaas.com/v1"

terminal:
  backend: "local"
  timeout: 180

agent:
  max_turns: 90
Then add your API key to ~/.hermes/.env: 
OPENAI_API_KEY="sk-..."

​
Provider options

The provider field accepts these values:
ProviderDescription
autoAuto-detect (requires direct provider credentials)
customAny OpenAI-compatible endpoint (set base_url)
openrouterOpenRouter multi-provider routing
lmstudioLM Studio local server
ollama, vllm, llamacppAliases for custom
Setting provider: "custom" with a base_url routes all requests through ClearMaas.

​
Using the Anthropic protocol

For native Anthropic requests (bypassing translation), use the bare host URL: 
model:
  default: "anthropic/claude-opus-4.6"
  provider: "custom"
  base_url: "https://api.clearmaas.com"
The key difference: ClearMaas’s Anthropic path (https://api.clearmaas.com) expects the SDK to append /v1/messages, while the OpenAI-compatible path (https://api.clearmaas.com/v1) is the complete endpoint.

​
Interactive setup

If you prefer a guided setup: 
hermes model
This interactive command walks through provider selection. Choose custom and enter:
  • Base URL: https://api.clearmaas.com/v1 (OpenAI-compatible) or https://api.clearmaas.com (Anthropic-compatible)
  • API Key: sk-...

​
Environment variable override

Override settings at runtime: 
export OPENAI_API_KEY="sk-..."
export OPENAI_BASE_URL="https://api.clearmaas.com/v1"
hermes
Environment variables take highest precedence, followed by CLI arguments, config.yaml, then .env.

​
Choosing models

Set the default model in config.yaml: 
model:
  default: "openai/gpt-4o-mini"
  provider: "custom"
  base_url: "https://api.clearmaas.com/v1"
Use any ClearMaas model prefix:
PrefixExample
openai/openai/gpt-4o
anthropic/anthropic/claude-sonnet-4.6
google/google/gemini-2.5-pro
deepseek/deepseek/deepseek-chat
grok/grok/grok-4-fast-reasoning
qwen/qwen/qwen3.6-plus

​
Config migration

After updating Hermes Agent, migrate your config to include new options: 
hermes config migrate

​
Verification

export OPENAI_API_KEY="sk-..."
export OPENAI_BASE_URL="https://api.clearmaas.com/v1"
hermes
# Type a question and verify the response
Check your ClearMaas dashboard to confirm requests are routed.

​
See also