Different claims have different shapes, and each shape has a tool that grips it. "If n is a multiple of 6, then n is even" hands you something to work with: n=6k, so n=2(3k), and forward reasoning finishes it in one line. That is a direct proof. "There is no largest prime" gives you nothing to start from, but its opposite, "there is a largest prime p", is concrete enough to attack, so contradiction fits. "2n≥n+1 for every integer n≥1" is a ladder of claims, each built from the last, which is the shape induction is made for.
Two more tools handle shapes the first three don't. The contrapositive replaces "if n2 is odd, then n is odd" with the equivalent "if n is even, then n2 is even", which has a usable starting point. Proof by cases splits a claim into pieces you can handle one at a time: to show that n2 leaves remainder 0 or 1 when divided by 4, take n=2k, giving 4k2, and n=2k+1, giving 4(k2+k)+1.
The technique is a plan, not the proof. Saying "use induction" earns nothing until the base case and the step are written out. Surface features can mislead, too: a letter n does not mean induction. "If n is even, then n2 is even" is about one arbitrary n, with no chain from case to case, so a direct proof is the natural fit. Choose by what the claim says, not by which letters it uses.
Think of it like this. Proof techniques are keys on a ring, and the claim is the lock. You don't try the same key in every lock; you look at the keyhole first: an if-then keyhole, a "there is no" keyhole, a "for every n" keyhole. Unlike real locks, a claim sometimes opens with two different keys, and then the shorter proof wins.
Core idea. Proof techniques are the menu of standard strategies — direct, contradiction, contrapositive, induction, cases — and the skill is choosing the one that fits the statement you must establish.