Zero2Repo：编码智能体能否从零构建代码仓库？

AI 摘要

研究团队发布 Zero2Repo 基准，用于评估编码智能体从零构建完整代码仓库的能力。任务由真实开源项目自动转换为需求文档、接口契约和隐藏验收测试，覆盖 Python、TypeScript、Go、C++ 共 11 个任务。评估显示最强智能体仅解决 10 个任务，且失败提交仍通过 90–99% 的隐藏测试，失败多源于规范中的低频规则或单一遗漏，而非缺少子系统。

正文节选

https://github.com/OpenEdgeHQ/Zero2Repo \homepagehttps://zero2repo.ai/ Zero2Repo: Can Coding Agents Build Repositories from Scratch? Abstract Coding agents are increasingly asked to build software rather than patch it, yet benchmarks for from-scratch repository construction are mostly limited to a single language and depend on manually curated tasks. We introduce Zero2Repo, a benchmark in which an agent receives a product requirements document, an interface contract, and an empty workspace, and

发布时间：2026-10-01 12:00

抓取时间：2026-10-01 13:31

来源机构：arXiv