How it works

intent-link watches how someone moves through a page and estimates which visible target they are approaching. When the signal is strong enough, it calls that target's onIntent function.

1 · Observe movement

On desktop, the engine reads pointer movement. On touch devices, it reads scrolling. It starts only when the user interacts and settles again when movement stops.

2 · Estimate the target

The engine compares the user's movement with the visible intent targets. Targets outside the viewport are excluded from active calculations.

3 · Run your action

When one target becomes likely enough, its onIntent callback runs once. It can run again after the user moves away and later approaches the target again.

Desktop and mobile

• Desktop prediction follows pointer movement in two dimensions.
• Mobile prediction follows vertical scroll movement. The first scroll activates it.
• Hidden and off-screen targets are ignored, including elements hidden by responsive CSS.

The physics idea

Imagine each target as a small gravity well. Movement creates kinetic energy, while distance from a target creates potential energy. A fast pointer or scroll still has motion left; a slow movement close to a target looks like it is settling there. The engine uses a thermodynamics-style rule that treats these lower-energy destinations as more likely.

Browser movement data contains tiny jumps and mistakes. A Kalman filter smooths that noise before the physics calculation. In simple terms, it helps the engine separate deliberate movement from shaky measurements. This all stays inside the library; applications only receive onIntent.