{
  "added_at": "2026-10-16",
  "arxiv_id": "2608.11469",
  "arxiv_version_reviewed": "v2",
  "authors": [
    "Jeremy Spence",
    "Nicholas Assaderaghi",
    "Feng Xiao",
    "Jinhao Zhu",
    "Nikil Ravi",
    "Xiangyu Qi",
    "Matthew Jagielski",
    "Raluca Ada Popa",
    "Eric Wallace",
    "Guannan Wei",
    "Yangruibo Ding",
    "Zhuo Zhang"
  ],
  "bibtex_url": "https://awesomerea.org/papers/citations/paper-sre-bench.bib",
  "code_url": null,
  "date_basis": "arXiv v1 submission; summary and author list use revised v2",
  "date_basis_zh": "arXiv v1 首次提交日期；摘要与作者名单依据修订版 v2。",
  "date_precision": "day",
  "description_en": "Introduces SRE-Bench, with privately authored programs and protected binary variants, to study agent recovery of program semantics while reducing training-data contamination.",
  "description_zh": "提出 SRE-Bench，使用私有编写的程序及带保护机制的二进制变体，研究智能体恢复程序语义的能力，并减少训练数据污染。",
  "doi": null,
  "entry_url": "https://awesomerea.org/papers/paper-sre-bench/",
  "entry_url_zh": "https://awesomerea.org/zh/papers/paper-sre-bench/",
  "evidence_urls": [
    "https://arxiv.org/abs/2608.11469",
    "https://arxiv.org/html/2608.11469v2"
  ],
  "first_arxiv_date": "2026-08-11",
  "id": "paper-sre-bench",
  "json_url": "https://awesomerea.org/papers/paper-sre-bench/record.json",
  "kind": "research",
  "markdown_url": "https://awesomerea.org/papers/paper-sre-bench/index.md",
  "markdown_url_zh": "https://awesomerea.org/zh/papers/paper-sre-bench/index.md",
  "metadata_notes": "Use the 12-author v2 list, revised 2026-10-01, rather than the older eight-author v1 search snippets. No conference venue or own public code repository was verified. AgentRE-Bench GitHub links in the bibliography refer to another benchmark.",
  "metadata_notes_zh": "使用 2026-10-01 修订的 v2 十二位作者名单，而非旧 v1 搜索摘要中的八位作者。未确认会议发表场所或论文自己的公开代码仓库。参考文献中的 AgentRE-Bench GitHub 链接属于另一项基准。",
  "pdf_url": "https://arxiv.org/pdf/2608.11469",
  "project_url": null,
  "publication_date": "2026-08-11",
  "publication_status": "preprint",
  "record_type": "paper",
  "runtime_tested": false,
  "scope_limitations_en": "The revised study reports 19 programs, 262 binary instances, and 1,572 graded tasks. Binaries and task instances are not publicly released; access is through an evaluation service. Its contamination-control claim is the authors' methodology, not an independent guarantee.",
  "scope_limitations_zh": "修订版研究报告 19 个程序、262 个二进制实例和 1,572 个评分任务。二进制与任务实例未公开发布，只能通过评估服务访问。污染控制是作者的方法论主张，并非独立保证。",
  "targets": [
    "binary"
  ],
  "title": "The Next Challenge for Agentic Cybersecurity: A Realistic, Contamination-Free Reverse Engineering Benchmark",
  "title_zh": "智能体网络安全的下一项挑战：真实场景且无训练数据污染的逆向工程基准",
  "url": "https://arxiv.org/abs/2608.11469",
  "venue": "arXiv",
  "verification": "source-reviewed",
  "verified_at": "2026-10-16",
  "why_fits_en": "Adds realistic-scale and contamination-aware evaluation context for agent-assisted reverse engineering without mislabeling a private corpus as open data.",
  "why_fits_zh": "为智能体辅助逆向补充真实规模和数据污染控制的评估背景，同时明确私有语料的访问限制。",
  "year": 2026
}