> For the complete documentation index, see [llms.txt](https://docs.skyone.cloud/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.skyone.cloud/english/admin/guardrails.md).

# Guardrails

**Guardrails** prevent data leaks in the workspace by detecting secrets, personal data, and prohibited content in conversations and uploads.

Here are the available features:

### Browse the tabs <a href="#browse-the-tabs" id="browse-the-tabs"></a>

* **Guardrails:** the list of available detectors and their settings.
* **Alerts:** incidents detected by guardrails. The number next to the tab shows unread alerts.

### Available guardrails <a href="#available-guardrails" id="available-guardrails"></a>

* **API keys:** known formats (AWS, OpenAI, GitHub, Google, Slack), JWTs, and PEM private keys.
* **Passwords and credentials:** explicit passwords, connection strings, and authorization headers.
* **Entropy-based secrets:** high-entropy tokens with no known format.
* **Personal data (LGPD):** Brazilian CPF and CNPJ tax IDs and payment cards, with real validation; phone numbers and PIX keys (Brazil's instant payment system) are optional.
* **File extensions:** blocks uploads by file extension (e.g., .env, .pem, .key).
* **File contents:** applies the text detectors to the contents of uploaded files.
* **Prohibited terms:** a workspace list of customer names, code names, and off-limits topics.
* **Sensitive information (AI):** AI-assisted detection of sensitive information.

{% hint style="info" %}
The LGPD is Brazil's General Data Protection Law. If your company operates in other countries, add custom guardrails for local ID formats, such as US Social Security numbers.
{% endhint %}

### Turn on and configure a guardrail <a href="#turn-on-and-configure-a-guardrail" id="turn-on-and-configure-a-guardrail"></a>

To turn on a guardrail, follow these steps:

1. In the **Guardrails** tab, turn on the toggle next to the guardrail.
2. Click **Configure** to adjust its behavior.
3. Save your changes.

The badge next to the name shows the configured action mode:

* **Audit only:** logs the incident without interrupting the user.
* **Notify and block:** alerts the user and stops the content from being sent.

{% hint style="info" %}
**Important:** start with **Audit only** to gauge how often incidents occur before you block content.
{% endhint %}

### Create a custom guardrail <a href="#create-a-custom-guardrail" id="create-a-custom-guardrail"></a>

To detect a pattern specific to your company, follow these steps:

1. At the top of the page, click **+ Create guardrail**.
2. In the **Name** field, enter a name for the guardrail.
3. In the **Regular expression** field, enter the pattern to detect. Example: `EMP-\d{6}` to detect employee IDs in the format EMP-123456.
4. Click **Save**.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.skyone.cloud/english/admin/guardrails.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
