Metabolism is driven by an intricate network of biochemical reactions that convert nutrients into energy and biomass, sustaining all aspects of life. The wiring of metabolic flux through these reactions, or the collection of their reaction rates, defines the metabolic status of living organisms. Knowing how flux is wired, and how it rewires when perturbed, is fundamental to understanding growth, development, homeostasis, and pathogenesis. However, our knowledge of metabolic wiring and rewiring remains highly fragmented, with systems-level principles largely unknown. This thesis presents a body of work that collectively establishes the first systems-level characterization of metabolic wiring and rewiring, using Caenorhabditis elegans as a model organism. Experimentally, we develop Worm Perturb-Seq, a massively parallel RNAi and RNA-seq technology, to generate a transcriptomic compendium of systematic metabolic gene perturbations. Computationally, we create metabolic network modeling methods to translate gene expression data into insights on both metabolic wiring and rewiring. Combining these two, we uncover the systems-level design principles of metabolic rewiring, which we named the Compensation/Repression model. Additionally, we reconstruct the first whole-network flux wiring map of an animal, leading to many surprising discoveries validated by isotope tracing. These include a non-canonical central carbon metabolism where C. elegans breaks down ribose from dietary RNA via a cyclic pentose phosphate pathway and uses dietary amino acids as a primary energy source. Together, this thesis reveals systems-level principles of metabolic wiring and rewiring in an animal, achieved through a unique genomic lens that combines high-throughput genomics with metabolic network modeling.