# pstack 0.15.0 发布：token 使用效率提升 3% 至 11%

- 来源：lauren (@poteto)
- 发布时间：2026-09-09 01:42
- AIHOT 分数：33
- AIHOT 链接：https://aihot.news/items/cmtszj1rh04g8ro5whyh83f6b
- 原文链接：https://x.com/poteto/status/2097380152703615396

## AI 摘要

pstack 0.15.0 发布，主要改进为 token 使用效率提升，根据会话中加载的 skills 数量可节省 3% 至 11%，并清理了多余分号、破折号及冗长表述。新增“攻击前提”和“测试行为而非实现”两条原则，以减少无用测试。使用 /poteto-mode 时这些原则会自动加载，也可通过 /principle-... 按需调用。

## 正文

pstack 0.15.0 is here! this release contains mainly token usage improvements (between 3 to 11% savings depending on how many skills are loaded by pstack in your session), removing some stray semicolons and emdashes so pstack skills aren't a bad influence, and lots of mannered prose removals (thanks Anthropic for the guide!), along with 2 new principles:

• attack the premise: inspired by @VictorTaelin's recent experience with agents never questioning the premise
• test behavior, not implementation: helps reduce the number of useless tests agents write

if you use /poteto-mode these principles should just be loaded automatically when needed, but you can always invoke on demand as well by autocompleting /principle-...

https://x.ai/bot/plugin/9717366