Hi @LenaLapidot,
No problem!
You can use both F and R reads for each sample or you can just import one of them, usually the forward reads have better quality than the reverse so if you were choosing only one, then the forward reads are probably your best bet. That being said, I would recommend starting with both F and R, referred to as paired-end, and follow the pipeline for those.
And yes you would have to build your manifest file manually unless you make some custom code for making this file automatically. For example, see one approach used for automatic manifest file making using R posted here.
Overall, if this is your first go at analyzing this type of data I would strongly recommend doing some additional readings first to understand the process otherwise it might get a bit overwhelming. For example, the qiime2 for dummies tutorial is a good start or even more comprehensive and theoretical is this documentation.. Good luck!