Maybe if the original question didn't mention them?
In comments also.
I will be grateful.
A reputation barrier wouldn't do, because sometimes a low rep user could still rightfully mention them, and high rep members might wrongfully do so.
I disagree, because this won't fix anything because there is nothing to fix. As I see it, recommending other techniques, frameworks or technologies can help in a lot of ways. Sometimes the questioner isn't aware of such possibilities and would be happy to use them.
I've never spent much time on C++, but the questions I do see are usually the result of failing to take proper advantage of the standard library. Many others are easily resolved with boost.
Thus, here's my counter proposal: Failing to articulate in your question why your problem is not solved by the standard library or boost results in being instantly banned from asking any question not directly related to 68k assembly.
I'm sympathetic! I'd upvote if I could. Not that I think there could or should be a global solution, but it gives people something to think about.
I believe most people would agree that it would be unacceptable to tell someone struggling with an algebra problem that they should just let WolframAlpha do their homework for them (while providing no other help or insight). Is suggesting boost and then peacing out a lesser version of the same thing?
I once found an example about splitting a up string using std::vector that really helped, and it became my first use of a vector (I hadn't heard of them): http://stackoverflow.com/a/236803/1736461
... although the poster was explicitly asking for elegance, so it might escape your ideal censure algorithm. And when you're doing certain kinds of homework (is there any other use for C++?) and you can't even use the STL, "use boost" is worse than useless.