> For the complete documentation index, see [llms.txt](https://doc.duaer.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.duaer.com/Duaer-Organizations/executions/execution-data-redaction.md).

# Redact execution data in Duaer

In Duaer, control the visibility of execution data to protect sensitive information and meet compliance requirements.
Execution data redaction hides the input and output stored on an execution. Status, timing, and node names stay visible. This is not encryption at rest, and it does not stop a node from passing data onward.

The instance setting requires a plan that includes data redaction. You can configure it per workflow, or enforce it instance-wide.

## Why use execution data redaction <a href="#why-use-execution-data-redaction" id="why-use-execution-data-redaction"></a>

Turning execution history off used to be the only option, and then you also lost success and failure. Redaction keeps monitoring while hiding the payload.

## Configure redaction settings <a href="#configure-redaction-settings" id="configure-redaction-settings"></a>

You need permission to enable or disable data redaction. Instance owners and project admins have that by default.

1. Open the workflow.
2. Select the three dots in the upper-right corner, then Settings.
3. Find Redact production execution data and Redact manual execution data.
4. For each, choose Default - Do not redact or Redact, then Save.

Production runs include webhooks, schedules, and other triggers while the workflow is active. Manual runs are those you start with Execute in the editor. You can hide manual runs only when production runs are hidden too. When the workflow uses end-user credentials, production execution data is always hidden.

## Instance-level enforcement <a href="#instance-level-enforcement" id="instance-level-enforcement"></a>

Instance owners and admins can open Settings → Security → Data redaction, turn on Enforce data redaction, and choose a scope:

- Production executions (recommended): every production run is hidden. Manual runs still follow each workflow’s own setting.
- Manual and production executions: both are hidden.
- Manual executions: only manual runs are enforced.

Enforcement applies when someone reads execution data. One workflow’s settings cannot be looser than the floor; the switches lock, and the public API rejects a looser policy. A workflow can still be stricter. New workflows start at the floor. Existing stored settings are not rewritten; if you later turn enforcement off, they return to those settings.

## What gets redacted <a href="#what-gets-redacted" id="what-gets-redacted"></a>

In the execution viewer, data tables become Data redacted. You still see node names, success or failure, timing, and structure.

- Each node’s input and output json is replaced with an empty object.
- Files and images in binary are removed.
- Fields marked sensitive (sensitiveOutputFields) stay hidden, including from someone allowed to reveal data.
- Error detail keeps the error type and, for API errors, the HTTP status code.
- A run that calls sub-workflows is redacted too when any of those sub-workflows has redaction on. If a sub-workflow’s output is not sensitive, turn off Redact the executions of Workflows that call this one in its workflow settings, and its callers are no longer redacted because of it.

## Reveal redacted data <a href="#reveal-redacted-data" id="reveal-redacted-data"></a>

Someone with Reveal execution data (execution:reveal) can reveal one execution. Instance owners and admins have this by default.

1. Open the execution.
2. Select Reveal data in the redacted area.
3. Confirm the dialog: the view is written to the audit log, and you should reveal only when you need to.
4. After you confirm, that execution’s data is visible in the current session.

Reveal is denied for executions that used dynamic credentials. A redacted execution also cannot be pinned back into the editor. The audit log records successful reveals, denied reveals, and changes to instance enforcement.

## What redaction does not cover <a href="#what-redaction-doesnt-cover" id="what-redaction-doesnt-cover"></a>

- console.log in a Code node is not redacted.
- Data still moves between nodes. Redaction only changes what a person sees in the execution viewer.
- A webhook response returned to the caller is the raw body.
- Outbound authentication headers are not rewritten.
- You cannot redact individual fields, except fields a node declares as sensitive.
- Execution data in the database is not rewritten or encrypted.
- Instance enforcement does not travel with source control to another instance.
## Questions

### What does redaction hide, and what stays visible?

In Duaer, input and output json become empty objects, binary is removed, and errors keep only type plus HTTP status for API errors. Node names, success or failure, timing, and structure stay visible. Declared sensitive fields stay hidden.

### How does per-organization redaction differ from instance enforcement?

In Duaer, one workflow can set production and manual redaction in Settings. An instance can enforce a floor under Security. An organization cannot be looser than the floor, but it can be stricter.

### What does redaction not cover?

In Duaer, it does not cover console.log, data still passed between nodes, raw webhook responses, or outbound auth headers. It is not per-field (except declared sensitive fields), does not encrypt the database, and enforcement does not travel with source control.

## Related

- [Customize executions data in Duaer](https://doc.duaer.com/Duaer-Organizations/executions/custom-executions-data.md)
- [Understand Duaer’s data structure](https://doc.duaer.com/data/data-structure.md)
- [Work with files and images in Duaer](https://doc.duaer.com/data/binary-data.md)

