What if gravity is not a force or a curvature, but a gradient in the rate at which time flows? We construct a complete scalar-tensor Lagrangian built on this idea. The fundamental dynamical variable is the local temporal flow rate τ(x) — the ratio of proper time to coordinate time — coupled to a tower of fractal texture fields with amplitudes scaled by powers of √2. From a single action, the theory recovers Newtonian gravity (fixing the kinetic parameter α = 2), the exact Schwarzschild metric (verified symbolically), modified Friedmann equations with dynamical dark energy, and an automatic cosmological bounce that forbids the Big Bang singularity. Dark energy arises naturally from frozen texture modes without an imposed cosmological constant, yielding an effective equation of state w₀ ≈ −1.0 with mild evolution at high redshift — compatible with Planck 2018 and within the parameter space explored by DESI 2024. The theory predicts log-periodic modulations in the CMB power spectrum with period ln√2, testable with current and near-future data. Gravitational waves propagate at exactly c, consistent with GW170817. The theory sits at the conformal coupling point (Brans-Dicke parameter ω = −1), which raises the ghost question. We address this in detail: the scalar mode is non-propagating in vacuum (the field equation is a constraint, not a wave equation), the potential provides an effective mass that decouples perturbations at sub-Hubble scales, and the exact Schwarzschild recovery ensures solar-system consistency (PPN parameter γ = 1 identically). A complete non-perturbative ghost analysis remains an open problem, flagged explicitly. The exponent p = √2 in the potential τ^p is not imposed but selected: it is the unique value in the stability interval (1,2) satisfying the self-consistency condition α = p². Whether this algebraic coincidence reflects a deeper principle is left as an open question. The theory has comparable parametric complexity to ΛCDM (8 parameters vs 6) while providing a dynamical origin for dark energy, a singularity-free cosmology, and multiple falsifiable predictions with explicit criteria dated to 2035.