AlloyDesignClaw: Optimizing Castability and Mechanical Properties of Nickel‐Based Superalloys

ABSTRACT A rational design of cast superalloys that simultaneously achieves robust processability and superior mechanical performance remains a formidable challenge due to the high‐dimensional and strongly coupled composition space. Traditional data‐driven materials design approaches were generally lack in user‐friendliness, making it difficult for widespread engineering application. Here, we present AlloyDesignClaw, a multi‐agent LLM‐assisted alloy design framework that integrates CALPHAD‐based thermophysical calculations, surrogate modeling, SHAP analysis, castability‐oriented multi‐objective optimization, and mechanical performance–driven screening to accelerate composition discovery. Using the K4169 superalloy as a case, AlloyDesignClaw rapidly identifies an optimized composition range that simultaneously achieves excellent castability, high ultimate tensile strength, and moderate elongation, solely based on user‐specified requirements expressed in natural language. Moreover, explainable analysis reveals Ni, C, Cr, Fe, Nb, and Si as the key elements governing the castability of the superalloy. Our work establishes an interpretable, user‐friendly, and efficient framework for alloy design, substantially accelerating the discovery and optimization of high‐performance superalloys.