用约束机制提升代码代理的可监管性，降低人工审查成本。
Steerability via constraints: a substrate for scalable oversight of coding agents
- 通过访问控制、编码规范等底层约束管理代码代理
- 审查召回率从54.5%提升至90.9%，仅需约200行工具代码
- 适合关注安全与可维护性的开发团队使用
代码代理能力强大，但人工监督成为瓶颈。无约束的代理会引入安全风险，损害代码库可扩展性，并使人工审查成本不断上升。我们认为，过去几十年用于管理大型人类工程团队的方法——如访问控制、网络策略、由工具强制执行的严格编码规范——可直接应用于代码代理，且在令牌消耗上比近期的代理框架更高效。我们基于此原则构建了一个端到端系统，并开展了一项可控实验：由小型评审者（Gemma 4 e4b）检查包含11个植入后门的Python代码库。在引入约束基底和约200行`docs` CLI工具后，召回率从无约束情况下的54.5%提升至90.9%，且基底与工具的贡献相互独立。选择Python是因其默认保障最少，约束带来的收益最大；该原则亦适用于Rust等语言。
原文摘要 · Abstract (English)
Coding agents are capable; human oversight is the bottleneck. Unconstrained agents introduce security risks, erode codebase scalability, and make human review increasingly costly. We argue that the same methods used for decades to manage large human engineering teams: access control, network policies, strict coding conventions enforced by tooling; transfer directly to coding agents, and are cheaper (in token) than recent agentic scaffolding. We sketch a start-to-end system on this principle, and report a controlled experiment in scalable oversight: a small reviewer (Gemma 4 e4b) inspects a Python codebase containing 11 inserted backdoors. Recall rises from 54.5% (unconstrained, no tools) to 90.9% (constrained substrate plus a ~200-LoC `docs` CLI), with substrate and tools contributing independently. We choose Python deliberately: substrate-level oversight gains are largest where the language gives the fewest guarantees by default; the principles extend to languages like Rust.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。