The $B^+ \to K^+ \nu \bar \nu$ decay as a search for the QCD axion

We introduce a model-independent framework to reinterpret Belle II results using only public data, analytically reconstructing the mapping between true and reconstructed kinematic variables within the statistically dominant Inclusive Tagging Analysis. This enables rare-decay measurements to probe light invisible particles -- such as the QCD axion or axion-like particles, collectively denoted $a$ -- without relying on internal simulations. Applying the method to $B^+ \! \to \! K^+ \nu \bar\nu$ yields the strongest bound on the branching fraction for $B^+ \! \to \! K^+ a$, improving existing limits by about a factor of nine and constraining the axion's fundamental flavour-changing coupling to $b$ and $s$ quarks. The approach establishes $B^+ \! \to \! K^+ \nu \bar\nu$ as a dual probe -- simultaneously testing short-distance new physics and light invisible states, the two probes working independently to an excellent approximation -- and provides a general strategy for model-independent reinterpretation of collider data.