Source Code for GRAFT

Language-model agents plan, execute, and evaluate scientific workflows. However, the problem--method relationships guiding their choices remain implicit, so plausible outputs can violate the assumptions of the scientific problem. We introduce Graph Reduction to Adaptive Factored Trees (GRAFT), which makes this problem-to-method map explicit as an expandable probabilistic structure of admissible problems, methods, and dependencies. Because this structure follows scientific semantics, knowledge from documentation, papers, and domain experts can be encoded, while factorization keeps the structure tractable. Semantic fingerprints of encoded problems measure their similarity, so expertise and experience guide related problems. GRAFT-guided agents match or exceed expert baselines across physics-informed learning, numerical solvers, blood rheology, and formal mathematics, including Apollo Command Module hypersonic loading within $1.8\%$ of experiment, a certified regularizer for brain-flow reconstruction, and a spectral physics-informed architecture with a machine-checked error bound. The representation retains outcomes and incorporates new methods, so what transfers grows with every solved problem.