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

# Policies — Data

> Guardrails for secrets and personal data in agent traffic

|              |                                             |
| ------------ | ------------------------------------------- |
| **URL**      | `/policies/data-policies`                   |
| **Sidebar**  | Policies → Data                             |
| **Roles**    | Admin · Analytics Viewer, reading           |
| **Requires** | Data policies enabled for your organization |

Data policies protect the content moving through your agents — detecting secrets and personal data, and deciding what happens when they appear.

## What's on the page

The policy list, and the editor when you open or create one.

| Section                      | What you set                                                                                                                                                                                          |
| ---------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Guardrails Configuration** | Which guardrails apply and whether each one **Blocks** or **Audits**. Two are offered and no more: **Secrets**, always; and **PII**, only where it is enabled for your organization — contact support |
| **When**                     | How many detections in one request before the guardrail acts, from 1 to 10. Offered on both guardrails, but only the PII threshold takes effect                                                       |
| **Configure Entities**       | The specific entities to detect, such as categories of personal data                                                                                                                                  |
| **Enforcement Message**      | What the developer sees when the policy acts. Offered for **Block**, up to 500 characters. The editor's own note names Claude Code, Codex and Cursor; it also reaches Copilot and Augment Code        |

## Filters and controls

Scope a policy to user groups, or leave groups empty to apply it organization-wide. Groups are managed on **Users → User Groups**.

## What you can do

| Action                         | Result                                                         |
| ------------------------------ | -------------------------------------------------------------- |
| Create or open a policy        | Opens the editor with the sections above                       |
| Set an **Enforcement Message** | Defines what the developer reads at the moment the policy acts |

<Tip>
  The enforcement message is the developer's explanation at the moment they are stopped. Naming the policy and the next step keeps the interruption short.
</Tip>

## Messages you may see

| Message                 | What it means                                                                      |
| ----------------------- | ---------------------------------------------------------------------------------- |
| `Failed to save policy` | The policy was not saved. Your entries remain on screen, so saving again is enough |

## Related

| Page                                          | For                                           |
| --------------------------------------------- | --------------------------------------------- |
| [Analytics — Data](/dashboard/analytics/data) | What these guardrails matched                 |
| [Settings](/dashboard/settings)               | What happens when agents cannot reach Unbound |