Set your focus, based on the research question/concern. Out of the universe of possible things you could note, decide what subset will attract your focus. For instance, if you're interested in collaboration, focus events when people interact with each other or each others' artifacts. If you're studying self-mediation, look at how people arrange and use artifacts for themselves.
Survey the surroundings. Take a moment at the beginning of the session to get a sense of the workspace. How is it arranged? Where does the participant face? What resources are in arm's length? Out of the research focus, which of those resources are most important? Record: write a description, take photos or video, sketch.
Develop a shorthand. You'll be taking lots of notes - you probably won't even pause once you get in the flow - and using abbreviations and symbols will help you get the most of the session. For instance, I typically use the symbols > and < to indicate when the participant is communicating to someone else or being communicated with.
Commit to the observation. Is it better to take the time to write that exact quote or or quickly paraphrase it so you can write more field notes? Should you lean over and copy what the participant is typing on the screen or lean back and note the entire scene? You'll have to make many such decisions. Trust your instincts, commit, and don't second-guess yourself.
Be sensitive to similarities (repeated patterns) and differences (deviations). Patterns show you the general flow of the work; deviations indicate unusual conditions. The more observations you conduct, the easier it will be to see both of these.
Be sensitive to the totally expected and the totally unexpected. It's easy to see things that you don't expect, because they stick out - but it's also easy to dismiss them as anomalies. Note them. It's sometimes hard to see things that you totally expect to happen - but that's headline news as well, since it takes a lot of work behind the scenes for something to happen reliably. Note them too.