This is the official code repository for Anomaly-Driven Correction Discovery (ADCD), a physics-informed symbolic regression framework designed to mimic the evolutionary nature of scientific discovery. Rather than learning entire equations from scratch, ADCD takes a known classical physical law and seeks to discover the mathematical structure of the dimensionless correction term (Δ) that explains the discrepancy between classical predictions and anomalous experimental observations. ADCD v2.1.3 — submission-ready release containing comprehensive paper polish and audit fixes (resolving all LaTeX cross-references/labels, correcting ablation statistics to match bar data, disclosing the automated correction mode selection, formatting physical constant expressions, and adding PySR hall-of-fame footnotes). Tier B+ benchmark: 82.8% ± 7.7% mean structural recovery (5 seeds); PySR fair gap 77.8 pp at 5% noise; evaluation regimes disclosure. PyPI: pip install adcd==2.1.3. Paper PDF included as supplementary material. Key Features: Cascaded Physics Gates: Enforces physical constraints including AST complexity, dimensional homogeneity, transcendental argument guardrails, and asymptotic limits (ARC). JAX-Traced Parameter Optimization: Leverages automatic differentiation and JIT-compilation using L-BFGS-B parameter fitting. Noise Robustness: Integrates Bayesian Information Criterion (BIC) to prevent overfitting on noisy data up to 10% noise. Real Data Infrastructure: Built-in loaders and benchmarks for Mercury orbital precession, Hydrogen Lamb shift, Blackbody radiation, and Muon g-2 anomaly.