Cognite Toolkit deployment clears the dataset assignment on a Streamlit app

We deploy a Streamlit app using the Cognite Toolkit. The app’s YAML configuration includes a dataSetExternalId, but after deployment the app is no longer assigned to that dataset.

The dataset exists in each environment and can be assigned manually through the UI. The issue occurs in all three environments (Dev, QA, and Prod).

Streamlit configuration:

externalId: stapp-pi-extractor-tag-manager

name: PI Extractor Tag Manager

entrypoint: main.py

description: Manage Riglogger PI extractor tags — queue new tag requests and monitor sync status

published: true

dataSetExternalId: Riglogger_User

Dataset configuration:

externalId: Riglogger_User

name: Riglogger_User

description: Dataset for Riglogger user access.

writeProtected: false

metadata:

  origin: cognite-toolkit

Directory structure:

streamlit/
  stapp-pi-extractor-tag-manager.Streamlit.yaml
  stapp-pi-extractor-tag-manager/
    main.py
    requirements.txt

Steps:

1. Open the Streamlit app in CDF and assign it to the Riglogger_User dataset.
2. Deploy the app using the Cognite Toolkit.
3. Check the app’s dataset assignment in CDF.

After deployment, the dataset assignment has been cleared. The YAML still specifies dataSetExternalId: Riglogger_User, and the dataset is available to select manually.
 

Is this a known issue or limitation with Toolkit deployments of Streamlit apps? Is there additional configuration required to preserve the dataset assignment?

I’ve checked the documentation but haven’t found an explanation.