It's a common and understandable question, but they correctly get quickly closed as unclear what you're asking or too broad. In my personal opinion an answer explaining risk and safety would be very much appropriate for those questions - most importantly that there's no such thing as a hard line between safe and unsafe. So my proposal is to create an "artificial" canonical question asking about whether a very general scenario is safe and then a community answer that tries to get that point (and more as appropriate) over. Including examples what good and concrete questions about risk/risk-mitigation one could ask. Then we can close future such questions as duplicate, the asker gets a relevant answer and information on how to better approach their problem and ask questions.
I'm not sure what you envision the answer to the canonical question would be. Would it be a short course on risk assessment? If so, I can see someone reading it and saying well, that doesn't really answer my question and then asking a question which might be good, or might be too broad and/or lacking in detail. I think we should just deal with such questions as they come up until they become too burdensome, and I think that will not happen for quite some time.
Anytime you think you can write a good question that will add value to TGO you should write it.
There is nothing questionable about the scope here, so not sure how asking if you should write the question is helpful. If you write a good question it be welcomed, if you write a poor question, it will closed and stay closed.
Short answer, not a comment so detracters can downvote as appropriate.
I've seen the canonical question argument come up on other question and answer sites.
There are pros and cons of this approach. How about I list some of each.
Fewer questions to sort through
Easier to find answers to a question
Eliminates trolling that seeks to manipulate that particular question format
Enables the site to be more attractive to search engines
Discourages questions seeking knowledge on specific content
Hides the answer to a specific questions with a range of irrelevant
Loses information in the merging of similar questions
Can go horribly wrong where completely different questions are naively
For me, I suspect the cons are a stronger argument than the pros, however if these were addressed in the solution proposed for introduction of canonical questions, then the scheme could be made to work.