Here is Munkres' definition of a basis for a topology:
I wonder if it is assumed that when one talks about a basis for a topology, then the topology being considered is the topology generated by this basis? The reason I'm confused are these two theorems:
In the highlighted part, he uses that $\tau$ is the topology generated by the basis $\mathcal B$, but he doesn't mention this in hypothesis. So is it implicitly assumed that $\tau$ is the topology generated by $\mathcal B$?
Similarly,
The second parahraph of the prove says that we must prove that $\tau'=\tau$, but this wasn't claimed in the statement of the theorem.



I'll give you that this is confusing here.
You should read the definition at the top as "$\mathcal B$ is a basis for some topology on $X$ iff (1) and (2) are satisfied." The topology that it is a basis for is the one generated by $\mathcal B$ according to the next definition. (And I presume that they show this is in fact a topology shortly after.)
Then Lemma 13.1 shows that there is an equivalent definition for the topology generated by $\mathcal B$: the collection of arbitrary unions of sets in $\mathcal B.$ There is a third common equivalent definition: the topology generated by $\mathcal B$ is the intersection of all topologies $\mathcal T$ with $\mathcal B\subseteq \mathcal T.$
Here is a summary that may be less confusing: