Evaluating LLM-Based 0-to-1 Software Generation in End-to-End CLI Tool Scenarios
This addresses the need for better benchmarks in intent-driven development for AI researchers and practitioners, though it is incremental as it focuses on CLI tools.
The paper tackles the problem of evaluating LLM-based generation of complete software from scratch by introducing CLI-Tool-Bench, a benchmark that assesses 0-to-1 CLI tool creation without predefined scaffolds, revealing that top models achieve under 43% success.
Large Language Models (LLMs) are driving a shift towards intent-driven development, where agents build complete software from scratch. However, existing benchmarks fail to assess this 0-to-1 generation capability due to two limitations: reliance on predefined scaffolds that ignore repository structure planning, and rigid white-box unit testing that lacks end-to-end behavioral validation. To bridge this gap, we introduce CLI-Tool-Bench, a structure-agnostic benchmark for evaluating the ground-up generation of Command-Line Interface (CLI) tools. It features 100 diverse real-world repositories evaluated via a black-box differential testing framework. Agent-generated software is executed in sandboxes, comparing system side effects and terminal outputs against human-written oracles using multi-tiered equivalence metrics. Evaluating seven state-of-the-art LLMs, we reveal that top models achieve under 43% success, highlighting the ongoing challenge of 0-to-1 generation. Furthermore, higher token consumption does not guarantee better performance, and agents tend to generate monolithic code.