The more "s*t" you fix, the better you understand the toolset and the better developer you become by making better decisions.
The powerless feeling you have could be because a certain toolset is currently new to you.
Eg. I received a legacy project from a junior developer ( an important one), it took me 4 months to fix all ( and huge) issues. Now it takes me 5-10 minutes to identify an issue ( after improvements). It's a similar situation, but from a big internal project instead of a "small" library.
While your manager doesn't pay you to solve an issue from an external tool. He does pay you to solve the issue in the internal project. It's the same thing, but another perspective.
If you need to spend 2 days in that opensource project and you're sure that's the cause, so be it. A manager "only" cares about the end-result, not the road to it.
Managers tend to pick products that are most aggressively sold to them. Either by salespersons, or by public opinions.
As for the market for developers you either meet the market by developing skills in the libraries/frameworks that are in demand or you choose a different niche. You have 7 years experience so there must be certain areas that you are comfortable with.
Beware, of fad-driven development. The latest shiny new thing doesn't guarantee that it is applicable for every possible application.
I tend to simplify things as much as possible and using my own code. Doing PoC's to see if i can adopt a certain "thing".
So my solution for being "powerlessness" is owning the code i use and therefore I know what it does.
I use tools a lot, but also because of that, I'm able to fix issues fast and mostly the error is from my end.
For example, i had some slow query generation in EF a couple of years ago and this led to the EF team implementing a Like operator. - https://github.com/dotnet/ef6/issues/115
( I actually had to debug the library and give a reasonable explanation to the team. They found a solution, that they already considered. Which i think is pretty awesome of them! I use the like operator a lot by now)