Search papers, labs, and topics across Lattice.
The paper introduces CODESTRUCT, a framework that enables LLM-based code agents to interact with codebases through structured action spaces operating on AST entities rather than unstructured text. This is achieved by providing `readCode` and `editCode` functions for retrieving and modifying syntactic units in a syntax-validated manner. Experiments on SWE-Bench Verified and CodeAssistBench demonstrate that CODESTRUCT improves Pass@1 accuracy (up to 5.0%) and reduces token consumption (up to 38%), especially for models prone to generating invalid patches.
LLMs editing code are far more reliable and efficient when manipulating ASTs instead of raw text, slashing invalid patches and token costs.
LLM-based code agents treat repositories as unstructured text, applying edits through brittle string matching that frequently fails due to formatting drift or ambiguous patterns. We propose reframing the codebase as a structured action space where agents operate on named AST entities rather than text spans. Our framework, CODESTRUCT, provides readCode for retrieving complete syntactic units and editCode for applying syntax-validated transformations to semantic program elements. Evaluated on SWE-Bench Verified across six LLMs, CODESTRUCT improves Pass@1 accuracy by 1.2-5.0% while reducing token consumption by 12-38% for most models. Models that frequently fail to produce valid patches under text-based interfaces benefit most: GPT-5-nano improves by 20.8% as empty-patch failures drop from 46.6% to 7.2%. On CodeAssistBench, we observe consistent accuracy gains (+0.8-4.4%) with cost reductions up to 33%. Our results show that structure-aware interfaces offer a more reliable foundation for code agents.
Amazon Science