Transforming feature distributions to Gaussian/Normal

For anomaly detection: Would you say using a Box-Cox transformation of the data is a way to normalize the distribution of a given (skewed, non-normal) feature?

Yes, I would say so.