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

# Hermes Agent

> Connect OverControl to Hermes Agent using a custom endpoint, API key, and Lume model.

Connect OverControl to **Hermes Agent** and use Lume models directly inside your coding workflow.

Hermes Agent supports a custom OpenAI-compatible endpoint, so you can connect it to OverControl using your API key, the OverControl Base URL, and a Lume model ID.

<Note>
  This guide covers Hermes Agent in the terminal.
</Note>

***

## Before you start

You need:

* an [OverControl account](https://app.overcontrolgroup.com/register);
* an OverControl API key;
* a terminal;
* a Lume model ID.

<Card title="Open the API dashboard" icon="key" href="https://app.overcontrolgroup.com/dashboard" cta="Open dashboard">
  Create and manage the API key you will use with Hermes Agent.
</Card>

***

## 1. Install Hermes Agent

On macOS or Linux:

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

On Windows, open PowerShell and run:

```powershell theme={null}
iex (irm https://hermes-agent.nousresearch.com/install.ps1)
```

When the installer finishes, open a new terminal.

***

## 2. Choose a custom endpoint

Run:

```bash theme={null}
hermes model
```

Choose **Custom endpoint**.

***

## 3. Configure OverControl

Enter the following values:

| Setting | Value |
| - | - |
| **API base URL** | `https://api.overcontrolgroup.com/v1` |
| **API key** | Your OverControl API key |
| **Model name** | `lume-3.5` |
| **Context length** | `400000` |
| **Display name** | OverControl |

For most coding workflows, start with **Lume 3.5**. Lume 3.5 has a 400,000-token context window, so enter `400000` when Hermes asks for the context length.

```text theme={null}
Base URL: https://api.overcontrolgroup.com/v1
Model ID: lume-3.5
Context length: 400000
Display name: OverControl
```

<Warning>
  Keep your API key private. Never expose it in screenshots, logs, public configuration files, or source-control repositories.
</Warning>

Hermes saves the endpoint on your machine.

***

## 4. Verify the connection

Start Hermes:

```bash theme={null}
hermes
```

Send a simple task:

```text theme={null}
Explain the structure of this project and identify the main files I should understand first.
```

If Hermes returns a response using the selected Lume model, the integration is ready.

***

## Working with Hermes Agent

Once connected, Hermes Agent can use OverControl while helping you:

* understand unfamiliar code;
* explain files and functions;
* plan implementations;
* debug issues;
* review or refactor code;
* write tests;
* draft technical documentation;
* reason about architecture.

Hermes Agent can also inspect files, propose edits, and run commands as part of a task.

<Warning>
  Review file changes and terminal commands before approving them, especially when working with sensitive projects or production systems.
</Warning>

***

## Choose a model

For most coding workflows, start with **Lume 3.5**.

Use **Lume 3.5 Max** when the task needs more reasoning, for debugging, architecture, or a longer chain of steps.

```text theme={null}
lume-3.5-max
```

Run `hermes model` again, choose **Custom endpoint**, and enter that model name. Keep the same base URL, API key, display name OverControl, and context length of `400000`.

[Compare Lume models](/models)

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="The hermes command is not found">
    Open a new terminal and run the command again. On bash, reload the shell with `source ~/.bashrc`. On zsh, reload it with `source ~/.zshrc`.
  </Accordion>

  <Accordion title="Hermes says the API key is invalid">
    Check that your OverControl API key was copied correctly and has not been revoked.

    You can create or manage keys from the [API dashboard](https://app.overcontrolgroup.com/dashboard).
  </Accordion>

  <Accordion title="Hermes cannot connect">
    Confirm that **Custom endpoint** is selected and that the base URL is:

    ```text theme={null}
    https://api.overcontrolgroup.com/v1
    ```

    Also check your network connection and API key.
  </Accordion>

  <Accordion title="The model is not found">
    Make sure the model ID is entered exactly as documented.

    ```text theme={null}
    lume-3.5
    ```

    [View available models](/models)
  </Accordion>

  <Accordion title="The context window is too small">
    Run `hermes model`, choose **Custom endpoint**, and set the context length to:

    ```text theme={null}
    400000
    ```

    Lume 3.5 and Lume 3.5 Max both use a 400,000-token window.
  </Accordion>

  <Accordion title="Requests fail after working previously">
    Check your selected model, API key, available quota or balance, and API usage in the OverControl dashboard.
  </Accordion>
</AccordionGroup>

***

## Next steps

<CardGroup cols={3}>
  <Card title="AI coding tools" icon="code" href="/integrations/ai-coding-tools">
    Return to the AI coding tools overview.
  </Card>

  <Card title="Models" icon="brain" href="/models">
    Choose the Lume model for your coding workflow.
  </Card>

  <Card title="Integrations" icon="plug" href="/integrations">
    Explore all supported integrations.
  </Card>
</CardGroup>


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