> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scitix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Hermes Agent

> Connect Hermes Agent to ScitiX Model Inference from the terminal using a custom OpenAI-compatible endpoint.

[Hermes Agent](https://github.com/NousResearch/hermes-agent) is Nous Research's open-source, self-improving AI agent. It runs in the terminal and on messaging platforms such as Telegram, Discord, and Slack. This guide connects Hermes Agent to ScitiX Model Inference using its `custom` provider.

## Install Hermes Agent

Install Hermes Agent using the command from the [official installation guide](https://hermes-agent.nousresearch.com):

```bash theme={null}
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
```

## Configure the connection

Create an API key on the ScitiX [API Keys](https://console.scitix.ai/model-inference/api_keys) page.

This example uses `zai-org/glm-5.3`. Confirm its tool-calling support and test it with Hermes Agent before use. For other available models and pricing, see the [Models](https://console.scitix.ai/model-inference/models) page.

Open `~/.hermes/config.yaml` and add or update the `model` section. Preserve any other existing settings, and replace `<YOUR_API_KEY>` with your ScitiX API key.

> [!WARNING]
>
> Keep your API key private. Do not commit configuration files containing the key to a repository or share them in screenshots.

```yaml theme={null}
model:
  default: zai-org/glm-5.3
  provider: custom
  base_url: https://api.scitix.ai/model-api/v1
  api_key: <YOUR_API_KEY>
```

| Field | Description |
| - | - |
| `default` | ScitiX model ID. This example uses `zai-org/glm-5.3`; choose a model that supports tool calling. |
| `provider` | Set to `custom`, not `openai`, to use an OpenAI-compatible endpoint such as ScitiX. |
| `base_url` | Set to `https://api.scitix.ai/model-api/v1`. Include the `/v1` suffix. |
| `api_key` | Your ScitiX API key. |

Alternatively, run `hermes model` and select **Custom endpoint** to configure the connection interactively.

> [!NOTE]
>
> `OPENAI_BASE_URL` applies to the `openai-api` provider, not `custom`. For a custom endpoint, set `base_url` in `config.yaml` or use `hermes model`.

## Verify the connection

Send a one-shot prompt and save the usage details (verified with `Hermes Agent v0.21.1`):

```bash theme={null}
hermes -z "Reply with exactly: OK" --usage-file usage.json
```

Check that Hermes returns `OK`, then inspect the saved usage file:

```bash theme={null}
cat usage.json
```

The following excerpt shows the relevant fields from the example output:

```json theme={null}
{
  "model": "zai-org/glm-5.3",
  "provider": "custom",
  "api_calls": 1,
  "completed": true
}
```

The `model` and `provider` fields report the selections used for the run. In this example, `api_calls` records one API call, and `completed: true` indicates that the run completed.

Hermes may identify itself as "Hermes Agent" regardless of the underlying model. Check the usage file or `hermes status` for the model in use rather than relying on the reply text.

To switch models during an interactive session, enter `/model custom:zai-org/glm-5.3`.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.