# Configuring an OpenTelemetry Collector for Data Export

> Use this guide to configure a generic OpenTelemetry (OTEL) Collector as a destination for UiPath Data Export. This option works with any OTEL-compatible observability platform that accepts OTLP (OpenTelemetry Protocol) over HTTP, including Langsmith, Langfuse, and MLFlow.

Use this guide to configure a generic OpenTelemetry (OTEL) Collector as a destination for UiPath Data Export. This option works with any OTEL-compatible observability platform that accepts OTLP (OpenTelemetry Protocol) over HTTP, including Langsmith, Langfuse, and MLFlow.

## Prerequisites

- An OTLP-compatible HTTP endpoint.
- An API key or other authentication credentials required by your collector (if applicable).
- Organization Administrator access in UiPath Test Cloud.

## Configuring an OpenTelemetry Collector

1. In UiPath Test Cloud, go to **Admin** and select **Data Export**.
2. Select **+ Add**.
3. **Step 1 – Select tenant**: Choose the tenant for this export, then select **Next**.
4. **Step 2 – Select data sources**: Select **Maestro & Agents**, then select **Next**.
5. **Step 3 – Configure destination**: Select **OpenTelemetry Collector** as the connection type and enter the following:

   | Field | Description |
   |---|---|
   | **API Key** | The API key required by your OTEL collector, if applicable. |
   | **Endpoint URL** | The OTLP HTTP endpoint URL (for example, `https://api.example.com`). |

   Optionally, add **Custom Headers** to pass vendor-specific metadata:

   1. Select **Add Header**.
   2. Enter the **Header Name** and **Header Value**.
   3. Repeat for each header required. Remove a header using the delete icon.

   :::note
   Custom header values are encrypted at rest. **Test Connection** is not available for the OpenTelemetry Collector destination.
   :::

6. **Step 4 – Review and save**: Review the configuration, then select **Save**.

**Result:** The new configuration appears in the **Data Export** table. To verify the setup, trigger any agent run (debug, evaluation, or runtime) or Maestro run, then confirm that traces appear in your observability platform.