How to do Scientific Writing (reports, articles, thesis,...)

Favorite Approach:

(In general writing is non-linear, but constancy is the most common denominator)

Read a lot.
Start with a Review on the topic (A review is a scientific paper that goes into what has been published in the topic). As you start reading more paper, you will see an introduction forming, and it will look like a funnel.
Google scholar is your best friend as a starting point. Remember to setup your library links (https://sites.google.com/view/mendozagroup/codestutorials/getting-started/literature-search-with-google-scholar )
When reading papers, you can start reading the figures, tables, and then read the full thing.
Mendeley, zotero for reference managing
For a paper, create the 4-5 figures you want to show to tell the story first and then you can start writing. (see also: https://sites.google.com/view/mendozagroup/codestutorials/advice/tips-for-creating-scientific-figures)

Just like above
Self-criticism, ask yourself, what is the purpose of this sentence?, is it related to the previous sentence?
This will help to improve your writing (or thesis), this might be one of the hardest parts to do. This is because you might have to remove some text to make more consistent or more to the point.
Read it out loud. You can chat with generative AI, by asking to do a critic. Also you can ask colleagues to read it and give you feedback.
For thesis, it could be an iterative process.

How to choose:

If you need open tools and LaTeX/BibTeX: Zotero or JabRef.
If your institution already licenses a manager: check whether you have free access to EndNote, Citavi or RefWorks.
If you live in Google Docs: Paperpile (or Zotero’s Google-Docs plug-in if you prefer open-source).
For collaboration with a diverse team: Zotero (free), EndNote (robust sharing), or Papers (slick but paid).
For heavy PDF annotation & discovery: Mendeley or ReadCube Papers.