A unified phenomenological model for gravitation across galaxies, wide binaries, dwarf spheroidals, and strong lenses

• New κ g t a p e r model reconciles dynamics across nine orders of magnitude in scale. • The framework is calibrated on disk galaxies and tested without re-tuning. • Explicit environment-dependence resolves the tension in wide-binary tests. • Model provides statistical preference over MOND and GR in strong-lensing data. • Environment-adaptive screening ensures consistency with Solar System constraints. Gravity is central to our understanding of nature, yet no single framework consistently reconciles all observational tests across astrophysical scales. Galaxy rotation curves and dwarf spheroidal dynamics show large acceleration discrepancies at low accelerations, while Solar System experiments, Solar-neighborhood wide binaries, and strong gravitational lensing tightly constrain departures from general relativity. Existing approaches, including dark matter and Modified Newtonian Dynamics (MOND), capture parts of this phenomenology but struggle to satisfy these constraints simultaneously without additional assumptions or tuning. Here we introduce an empirically calibrated, environment-adaptive gravitational kernel, κ g t a p e r , as a phenomenological rule in the weak-field regime. The effective response is allowed to vary with an external-field environment variable derived from the surrounding gravitational field. Calibrated solely on rotation curves of isolated disk galaxies, the framework is tested without re-tuning against 14 564 wide binaries in the Solar neighborhood, dwarf spheroidal satellites, and 153 strong-lensing systems. The model predicts screening in high-external-field environments, yielding dynamics consistent with general relativity for wide binaries and Solar System tests, while retaining a weak-field enhancement in low-external-field systems such as isolated galaxies and distant dwarf spheroidals. In strong-lensing data and across independent validation sets, κ g t a p e r provides better likelihood-based fit than both general relativity with baryons only and standard MOND formulations under the specified comparison framework. These results show that an explicitly environment-dependent phenomenological rule can quantitatively reconcile several long-standing, and previously conflicting, astrophysical tests using a single fixed set of parameters. While phenomenological and not yet derived from a covariant theory, the model provides a compact empirical target and quantitative constraints for future theory development, but it does not by itself identify a physical mechanism or rule out dark matter-based explanations.