Can we add parameters/secrets while pushing the kernels ?

[bleakprestiger]

Can we add parameters/secrets while pushing the kernels ?

[stevemessick]

Could you give us a little more detail? An example of what you want to do would be great.

[YuenSzeHong]

Could you give us a little more detail? An example of what you want to do would be great.

I guess he ment user secrets (like api keys that we dont want to expose)

[stevemessick]

No, I do not think we want to do that. I think it would introduce security issues. @rosbo Could you confirm?

The way to manage your API key is with the kaggle_secrets package. See:
https://www.kaggle.com/code/harupy/how-to-use-kaggle-api-as-a-python-library

This package is only available in Kaggle notebooks and is a secure way to set up user name and key.

[rosbo]

@stevemessick To clarify, kaggle_secrets is used to access secrets in a notebook (i.e. get the secret value).

from kaggle_secrets import UserSecretsClient

secret_value = UserSecretsClient().get_secret("YOUR_SECRET_NAME")

kaggle_secrets doesn't allow setting the value, it only reads the value. Currently, the value must be set from Notebook Editor > Add-ons > Secrets. Another thing to note is that secrets are set at your account level. Then, you can decide which secrets you want to attach to a specific notebook. If you update the value of a secret, the new value will be used by all notebooks using that secret.

@bleakprestiger Are you asking for an option to "attach" a given list of secrets automatically when calling kaggle kernels push?

Or are you asking for new methods in the CLI to set user secrets. Something along the lines of:

kaggle secrets set MY_API_KEY abcdefghijk123

Thank you