README.md

metadata

pretty_name: TestSuiteBench Reference Tests
license: other
viewer: false
tags:
  - code
  - software-testing
  - benchmark

TestSuiteBench Reference Tests

Reference test suites for the 165 tasks of TestSuiteBench, a benchmark of whether coding agents can write a test suite for a real CLI tool from its specification alone. The tests come from ProgramBench. They are used to validate the mutants and to score code-generation runs.

Usage

See the GitHub repository for how to use this dataset.

Citation

@misc{egashira2026testsuitebench,
      title={TestSuiteBench: Coding Agents Struggle to Generate Strong Test Suites from Specifications},
      author={Kazuki Egashira and Niels Mündler-Sasahara and Mark Niklas Müller and Martin Vechev},
      year={2026},
      eprint={2610.13065},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2610.13065},
}

The task instances are taken from ProgramBench. Please consider citing it as well:

@misc{yang2026programbench,
    title={ProgramBench: Can Language Models Rebuild Programs From Scratch?},
    author={John Yang and Kilian Lieret and Jeffrey Ma and Parth Thakkar and Dmitrii Pedchenko and Sten Sootla and Emily McMilin and Pengcheng Yin and Rui Hou and Gabriel Synnaeve and Diyi Yang and Ofir Press},
    year={2026},
    eprint={2605.03546},
    archivePrefix={arXiv},
    primaryClass={cs.SE},
    url={https://arxiv.org/abs/2605.03546},
}