Directional Anisotropy Gravity

General relativity and the Standard Model are each empirically successful, yet they rest on incompatible primitives: dynamical spacetime versus quantum fields on a fixed background. The usual response is to bridge them by adding structure on top. Directional Anisotropy Gravity (DAG) inverts the move: it asks what minimal discrete substrate any directed, observable reality must already presuppose, and whether established physics can be recovered from it with no further free input. The architectural input is eight postulates — six fixing a discrete, oriented, six-channel substrate with one Direction–Orbit Coupling Hamiltonian, one relational-energy prescription, and one open-system memory channel; plus two substrate-to-geometry bridges: a gravitational Born rule for the temporal lapse, and a site-local Regge edge map for the spatial ruler. From these commitments the paper derives: — a controlled general-relativistic limit, with an Einstein–Hilbert infrared action, two massless transverse-traceless gravitons, post-Newtonian parameters β_PPN = γ_PPN = 1, Nordtvedt η = 0, and the exponential static-vacuum metric in closed form for the strong field; — a universal matter-coupling theorem whose weak-equivalence-principle residual δa / a ≲ 10⁻²⁶ sits eleven orders below the current Eötvös / MICROSCOPE bound; — a one-parameter dark sector controlled by a single memory rate Γ_F: its inhomogeneous projection reproduces the baryonic Tully–Fisher law v_f⁴ = G M_b a_*, and its homogeneous projection scales the residual vacuum-energy density structurally as (H / M_Pl)², recovering the observed ρ_Λ / ρ_Planck ~ 10⁻¹²² up to an O(1) coefficient; — a structural attachment to the Standard Model gauge group S(U(3) × U(2)) ≅ [SU(3) × SU(2) × U(1)] / Z_6, with exactly three chiral families, a structurally fixed hypercharge lattice (integer charge quantisation on physical colour-singlet states, with overall lattice scale matched to Y(Q_L) = 1/6), and no light exotics. The one-rate dark sector has already been tested on the 149-galaxy SPARC sample: the mean activation-radius ratio sits at the upper edge of the predicted band ⟨r_t / r_AQUAL⟩ ∈ [7, 10], consistent with ε ≈ 10⁻² within the empirical uncertainty. DAG commits in advance to specific galactic memory effects, strong-field deviations, and null results in regimes where current experiments already enforce them. Three status terms are used deliberately throughout: derived (the result follows from the admitted substrate without new free input); structurally derived (the mechanism and term structure are closed, while a precision coefficient remains to be determined numerically); conditionally derived (the result follows after an explicitly stated closure). Where a quantity is fixed by matching to data, the text says so. Companion documents in the DAG public set:— Technical Supplement (DOI 10.5281/zenodo.20397317): full proofs of every theorem stated here, assumption ledger, and dependency tree.— Predictions, SPARC Verification, and Operational Tests (DOI 10.5281/zenodo.20397375): operational distillation for the observational researcher.— Programme sourcebook v12.2 (DOI 10.5281/zenodo.20071356): the full research programme, including ongoing work outside the demonstrated core.