It borrows YAML's indentation-based structure for nested objects and CSV's tabular format for uniform data rows, then optimizes both for token efficiency in LLM contexts. For deeply nested or ...