Overfitting model in validation dataset and test accuracy

Why do you conclude that if we have a test accuracy close to validation accuracy then the model have not overfit?
Should we fiddle with the hyperparameters to find the best one on the validation dataset? If the answer is Yes.
It doesn't make that the model overfitting on the validation dataset and then the validation accuracy that the model shows is for an overfitting model? I mean, is valid that reference us to that validation accuracy of the overfitting model?