KAJAL is an AI-assisted framework for inferring source-code grammars with large language models. It analyzes code examples to recover structural rules that support program understanding when formal grammar specifications are missing, incomplete, or difficult to maintain.
The project uses modular Python pipelines for prompt construction, grammar extraction, validation, and analysis. By connecting LLM reasoning with program-analysis workflows, KAJAL supports program comprehension, documentation, and downstream software engineering tasks across unfamiliar languages and codebases.