Skip to main content
Hermes Agent configured with Tinfoil
TL;DR — Configure Hermes Agent to use Tinfoil:
  • Install Hermes
  • Run hermes model and select Custom endpoint
  • Enter the base URL (https://inference.tinfoil.sh/v1/), your Tinfoil API key, and a model name (e.g., kimi-k2-5)
Verification Note: When using Hermes or other OpenAI-compatible clients, connection-time attestation verification is not performed automatically. However, all Tinfoil enclaves support audit-time verification through attestation transparency, creating an immutable audit trail. For applications requiring connection-time verification, use our official SDK clients. Learn more about the verification approaches.

​
Introduction

Hermes Agent is a self-improving AI assistant from Nous Research. This tutorial shows how to configure Hermes to route all inference requests through Tinfoil’s confidential computing enclaves.

​
Prerequisites

  • Tinfoil API Key: Get one at tinfoil.sh
  • Terminal: macOS, Linux, or WSL on Windows
You are billed for all Tinfoil Inference API usage. See Tinfoil pricing.
Never commit your API key to version control.

​
Installation

Follow the Hermes Agent installation guide to install Hermes on your system. Verify installation: 
hermes --version

​
Configure Tinfoil as the Provider

​
Step 1: Open Model Configuration

Run the interactive model selector: 
$ hermes model
Select provider:

  GitHub Copilot ACP (spawns `copilot --acp --stdio`)
  Google AI Studio (Gemini models — OpenAI-compatible endpoint)
  Z.AI / GLM (Zhipu AI direct API)
  Kimi / Moonshot (Moonshot AI direct API)
  MiniMax (global direct API)
  MiniMax China (domestic direct API)
  Kilo Code (Kilo Gateway API)
  OpenCode Zen (35+ curated models, pay-as-you-go)
  OpenCode Go (open models, $10/month subscription)
  AI Gateway (Vercel — 200+ models, pay-per-use)
  Alibaba Cloud / DashScope Coding (Qwen + multi-provider)
→ Custom endpoint (enter URL manually)
  Cancel

​
Step 2: Enter Tinfoil Configuration

Use the arrow keys to highlight Custom endpoint, then press Enter. You will be prompted for three values:
SettingValue
Base URLhttps://inference.tinfoil.sh/v1/
API keyYour Tinfoil API key from tinfoil.sh
Model nameWe recommend kimi-k2-5 (Kimi K2.5) or gemma4-31b (Gemma 4 31B). See the chat model catalog for all options.
Enter base URL: https://inference.tinfoil.sh/v1/
Enter API key (or press Enter to skip): <your-tinfoil-api-key>
Enter model name: kimi-k2-5

Configuration saved to ~/.hermes/config.yaml

​
Step 3: Verify Configuration

$ hermes doctor
✓ Config file exists
✓ Dependencies installed
✓ API connection successful
✓ Model kimi-k2-5 is accessible

​
Resources