If you are new to this, ask the editor of the journal for guidance. They probably have some prepared.
The minimum is to make a judgement about whether you think the paper should be published at all. If it is obviously sloppy or incorrect then you don't really need to do more than say that you think it can't be salvaged.
But, to serve both the journal and the authors you can, for reasonable papers, provide any advice that you think will (a) be required to make the paper come up to the journal's standards and (b) improve it for readers.
For a review of the literature you can point out that parts are missing if you notice that and possibly provide some pointers. But your job isn't to complete the paper.
This can include some minor suggestions on the writing and on the quality of the images. But you aren't expected to be a copy-editor. But if the language is terrible, you need to say that and, if possible, say how the author might go about getting it improved.
There are very rare cases, say in mathematics, in which the paper contains a poorly written but very important result. Then you, the editor, and the authors have the possibility of getting something important out, even if it requires a lot of work. For such things you can make more extensive suggestions if you have the time.