别反复提示：把智能体规则变成可执行检查

核心内容

Greg Pstrucha 以 Sentry 的调试智能体 Seer 为背景，讨论反复纠正编程智能体为何难以形成稳定行为。按发布方介绍，演讲主张把重复规则写入测试、类型约束和 lint 检查。

内容还涉及 API 与 OpenAPI 的一致性、技能文档检查，以及无法形式化的判断如何由书面规则和审核循环支持；同时提醒指标可能被优化行为钻空子。

Greg Pstrucha draws on Sentry’s debugging agent Seer to discuss why repeated corrections do not reliably stabilize coding-agent behaviour. The publisher describes codifying recurring rules as tests, type constraints and lint checks.

The talk also covers API/OpenAPI consistency, checks for skill documentation, written policies and review loops for qualitative judgments, and the risk of gaming metrics.

介绍依据 · 发布方说明 · 完整内容待核读

编辑导览，依据原始来源 ↗，整理于 2026-10-10。中英文均由本站概括整理。内容介绍与下方原话、主张的逐条核对分开标注。

智能体缺乏跨会话记忆，靠人工反复提示纠错既脆弱又低效；应对反复出现的编码偏差，应当让智能体根据历史 Review 自动生成确定性自定义 Lint 规则，并将定性经验沉淀为可执行的代码库策略审查循环。

原话

2逐字核对过；核对了说话人和位置

look at my GitHub reviews the feedback from the bots the feedback from my peers and try to codify which ones of those are actual high quality lint rules that are deterministic and will be able to capture issues that we otherwise reprompt all the time.

自动字幕 · 已核对Greg Pstrucha▶ 00:05:29

where the things that are writing code do not have any memory and are going to make the same repeated mistakes over and over again. And on the on the at the same time to write a new lter is very cheap and very easy. You can just have the agent do that

自动字幕 · 已核对Greg Pstrucha▶ 00:04:55