Hang a 2.0 kg bag from a spring scale and it reads 19.6 N. That reading is the tension: the string pulls up on the bag exactly as hard as gravity pulls down, because the bag is not accelerating. Now jerk the scale upward. While the bag speeds up at 2.0 m/s², the reading jumps to 23.6 N, because the string must hold the bag up and also speed it up: T=m(g+a)=2.0×11.8. Lower it with a downward acceleration and the reading drops below 19.6 N.
A light rope passes the same pull along its whole length, and at each end it pulls the attached object toward the rope's other end. If two people each pull on opposite ends of a rope with 100 N and nothing moves, the tension is 100 N, not 200 N: cut the rope anywhere and each piece pulls on the other with 100 N. Tie one end to a wall instead and the wall supplies the second 100 N; the tension is unchanged.
Contrast a rope with a rigid rod. A rod holding up a shelf bracket can push, so the force along it can point toward the object; that is compression, not tension. A rope can only pull. Push on a rope and it goes slack, and a slack rope has T=0. If a connector's force would have to point into the object to make the problem work, the connector cannot be a rope.
Think of it like this. A rope behaves like a line of people passing along a tug: each link pulls on its neighbours with the same force, so a pull at one end arrives undiminished at the other. The picture fails for a heavy rope, where links near the top must also hold up the rope hanging beneath them.
Core idea. A taut rope pulls each attached object along the rope with a size set by what that object is doing: equal to the weight it holds at rest, larger while the load speeds up upward, and zero once the rope goes slack.