> For the complete documentation index, see [llms.txt](https://ai-security-docs.akto.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ai-security-docs.akto.io/akto-argus-agentic-ai-security-for-homegrown-ai/agentic-ai-discovery/how-to/redact-sensitive-data.md).

# Redact sensitive data

Agent prompts and responses often contain PII and credentials. Automatic redaction lets your team review agent activity without exposing that data.

## Where Redaction Applies

| Where                  | What you see                                                                                                     |
| ---------------------- | ---------------------------------------------------------------------------------------------------------------- |
| **Guardrail Activity** | Data is always redacted, for everyone.                                                                           |
| **Traces**             | Admins see the full prompt, unredacted. Other users cannot open prompts. Redaction is available here on request. |

## How Redaction Works

Akto detects sensitive values in requests and responses and replaces each one with `[REDACTED]`. Only the sensitive values are replaced. The rest of the content stays readable, so you can still follow what the agent did.

```
... delta3 echo4 foxtrot5 golf6 [REDACTED] papa15 alpha16 bravo17 ...
```

## Guardrail Activity

In **Guardrail Activity**, the **Evidence** column shows the flagged content with sensitive values redacted. Open an event to view its **Request** and **Response**, where detected values are also redacted. This applies to all users, including admins.

Redacted values appear as `[REDACTED]`, or as a labeled marker such as `[SSN_REDACTED]` when Akto identifies the type of data.

<div data-with-frame="true"><figure><img src="https://1895868029-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkpBG2Nfd5ZbNmahr8W6H%2Fuploads%2Fgit-blob-42e28ff3d6f97f295d50f244e28dda20c6bb9323%2Fimage%20(251).png?alt=media" alt="" width="563"><figcaption></figcaption></figure></div>

## Traces

Prompts in **Traces** can contain sensitive data, so access depends on your role:

* **Admin**: you can open a trace and see the full prompt, without redaction.
* **All other users**: you cannot open prompts. Traces show a lock icon, and opening one shows "Please contact your administrator to get access."

<div data-with-frame="true"><figure><img src="https://1895868029-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkpBG2Nfd5ZbNmahr8W6H%2Fuploads%2Fgit-blob-267505c3ed17bb301bdbde7cd9227f6225980176%2Fredact-traces-non-admin.png?alt=media" alt="" width="563"><figcaption></figcaption></figure></div>

## Important Notes

* Redaction is automatic. You do not need to configure it.
* Admins can see full prompts in **Traces**, so give the admin role only to people who are allowed to view sensitive data.
* The same admin sees `[REDACTED]` in **Guardrail Activity** and the full prompt in **Traces**.