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

# Advanced agent settings

> General, AI, notifications, teams, canned responses, security, credit limits and variables for each agent.

# Advanced agent settings

Each agent has a **Settings** tab with these sections: **General**, **Notifications**, **AI**, **Teams**, **Canned responses**, **Security**, **Limits** and **Variables**. In the same menu you'll see **Satisfaction survey** and **Custom domain** with the **Coming soon** label: they aren't available yet.

To get there: `My agents → [your agent] → Settings`.

## General

* **Agent ID** — the agent's unique identifier, used when interacting with the [API](https://docs.chatnorris.ai/en/api-referencia/introduccion.md). It can't be changed; use **Copy** to copy it.
* The agent's **Name** and **Description**.
* **Timezone** — the time zone used by time-dependent features, including the AI: the agent knows the current day, date and time in this time zone, on all channels.
* **Danger zone** — **Delete agent** deletes the agent and its entire knowledge base. It asks for confirmation and can't be undone.

## AI

In the **AI** section you define how the agent replies:

* **Model** — the agent's AI model. Each model consumes a different number of credits per reply; see [AI models and credits](https://docs.chatnorris.ai/en/guias/modelos-ia-creditos.md).
* **Temperature** — from **Precise** to **Creative**: a low value gives more focused and consistent replies; a high one, more varied replies. Some models use a fixed temperature set by their provider, and this control doesn't apply to them.
* **Max tokens** — the maximum length of each reply.
* **Limit responses to the knowledge base** — when on, the agent answers only with its knowledge base and doesn't use general knowledge. When off, it can use general knowledge when the knowledge base has no relevant information.
* **System Prompt / Instructions** — the agent's instructions. With **Explore templates** you can apply a template based on the agent type; if the template has its own fields, you edit them in **Template fields** and apply them with **Apply to system prompt**.

## Teams

In **Teams** you create this agent's teams (name, description, color and members) to route handoffs to the right people. See [Teams per agent](https://docs.chatnorris.ai/en/guias/equipo.md#teams-per-agent).

## Security

Controls who can access this agent's web widget.

> [card] Domain whitelist
>
> List of domains where the web widget can load. Any site outside the list won't be able to show it. Format: `ejemplo.com`, without protocol or spaces.

> [card] Geo-blocking
>
> Blocks visitors from specific countries — the web widget is hidden for those countries.

> [card] Blocked users
>
> Blocks individual visitors by their `visitor ID` so they can't use the web widget.

> [card] Rate limiting
>
> Limit on the messages each visitor can send in a time window (minute, hour or day), with a **Custom error message** when it's exceeded. If you leave the messages field empty, the system's default limit is used.

> [!TIP]
>
> The domain whitelist and geo-blocking apply only to the web widget: they don't affect any messaging channel (WhatsApp, Instagram, Telegram, Messenger or Email).

## Variables

Variables let you inject dynamic data into the agent's system prompt.

### Custom variables

Create your own variables from `Settings → Variables → New variable`:

| Field | Description |
| - | - |
| **Name** | The variable's identifier (letters, numbers and underscores only). |
| **Type** | **Text**, **Number** or **Boolean**. |
| **Description** | What it's for — helps you remember it when writing the prompt. |
| **Default value** | The value it takes if no other is specified at runtime. |

### System variables

Besides the ones you create, ChatNorris exposes read-only system variables (marked with the **System** label in the table) that are filled in automatically based on the visitor and the conversation:

| Variable | What it contains |
| - | - |
| `contact_fname`, `contact_city`, `contact_state`, `contact_country`, `contact_postal_code`, `contact_address` | The visitor's location/contact details, when available. |
| `contact_company`, `contact_job_title`, `contact_industry`, `contact_website` | The contact's company details, when available. |
| `webpage_title`, `webpage_url` | Title and URL of the page where the widget is — web widget only. |
| `user_id`, `chat_id` | Unique identifiers of the visitor and of the current conversation. |
| `conversation_content` | Chat transcript (last 50 messages). |

## Canned responses

Text shortcuts so a human agent can reply faster during a handoff, without having to type the same replies again.

**1. Go to Settings → Canned responses**

You'll see the list of canned responses for this agent.

**2. Create a response**

Click **Create canned response** and fill in **Title** (to identify it in the list), an optional **Shortcut** (text shortcut, for example `/saludo`) and the **Body** (the message that will be sent).

**3. Use it during a handoff**

From the live chat, select the canned response to insert it without typing it again.

## Credit limits

On top of your plan's general AI credit pool, you can set a monthly credit cap **per agent** — useful if you have several agents and want to split usage among them.

* Turn on the switch to enable a specific monthly cap for this agent.
* **Monthly limit** — the maximum number of AI credits the agent can consume in the month.
* The month's current usage (actual usage, not estimated) and the counter's reset date are shown on the same screen.

> [!NOTE]
>
> This limit is in addition to your general plan's credit pool — it's for controlling a specific agent's usage, not for extending it beyond your plan.

## Agent notifications

From `Settings → Notifications` you can change, for this agent only, the channels (**Email**, **Bell**, **Push**) and the email recipients for its conversation events:

| Event | When it alerts |
| - | - |
| **Handoff requested** | A visitor asks to talk to a person. |
| **Unattended handoff** | Nobody replied to a handoff within the Human Handoff action's first reply target. |
| **New lead** | A visitor leaves their contact details. |
| **Unanswered question** | The visitor's last message has gone unanswered for a while. |
| **New conversation** | A new conversation starts with this agent. |

Each event can stay on **Inherit**, to use the organization's settings, or have its own values for this agent. Credit events (usage and credits exhausted) are configured only at the organization level. See the full detail of events, channels and recipients in [Notifications](https://docs.chatnorris.ai/en/guias/notificaciones.md).

## Frequently asked questions

> [accordion] Does the domain whitelist affect WhatsApp, Instagram or other channels?
>
> No. It only applies to the web widget. Messaging channels are managed separately from the Deploy tab.

> [accordion] Can I use custom variables together with system variables in the same prompt?
>
> Yes. Both are referenced the same way in the agent's system prompt.

> [accordion] Are canned responses shared between agents?
>
> No, each agent has its own list of canned responses.

> [accordion] What happens if an agent reaches its monthly credit limit?
>
> The agent stops replying until the counter resets the following month, even if your general plan still has quota available. You can turn off the limit at any time.


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