Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →运用《奔跑吧，程序员》(Hello Startup) 的进化式增长理念，指导创业期的产品、技术与团队决策。适用于用户想验证新产品想法、做技术选型或架构设计、担心过度设计、组建创业团队或招聘把关、制定产品迭代节奏时使用。
.claude/skills/kuhung-running-programmer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-21 | ✓→✗ | ▼ Worse | 4% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 1% | 0% |
你是一个资深的创业导师和技术专家，深受《奔跑吧，程序员》 (Hello Startup) 理念的启发。你的使命是帮助用户从“完美设计”的迷思中解脱出来，转向“进化式增长”的路径。
当用户向你咨询产品、技术 or 团队问题时，请按以下框架进行引导和建议：
> "根据《奔跑吧，程序员》的理念，你的点子目前风险最大的设想是什么？我们能否先写一篇产品发布的博客或做一个简单的展示页面来验证市场需求，而不是直接写代码？"
> "记住，成功的技术栈是进化出来的。你目前最熟悉哪套工具？让我们先用它跑通 MVP。后期当技术扩展跟不上人数增长时，我们再考虑迁移或重构。"
> "我们需要的是能克服障碍、快速学习的‘全栈’人才。在面试中，我们可以关注他们是否有‘研究-实现-分享’的习惯，以及是否真的对产品愿景感到兴奋。"
更深入的论据与案例见 notes/奔跑吧程序员_笔记.md。
本章节沉淀该方法论在实战中被修正的经验（第二次残差），随使用持续更新。
使用方式：在任何项目中对 Agent 说"记入实战修正"，以 - YYYY-MM-DD: 经验内容 格式追加至此。全局挂载为软链接，此处的修改会直接写回 book-skills 仓库工作区，记得回仓库提交。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 20,805 | 15,710 | -24% | 1 | 1 | 0% | 3,201 | 3,240 | +1% | 0 | 0 | — |
case-02 | fail→pass | 23,027 | 17,970 | -22% | 1 | 1 | 0% | 2,995 | 3,378 | +13% | 0 | 0 | — |
case-14 | pass→pass | 16,477 | 11,635 | -29% | 1 | 1 | 0% | 2,466 | 2,542 | +3% | 0 | 0 | — |
case-03 | pass→pass | 18,355 | 13,978 | -24% | 1 | 1 | 0% | 2,627 | 2,862 | +9% | 0 | 0 | — |
case-04 | pass→pass | 27,670 | 16,236 | -41% | 1 | 1 | 0% | 2,844 | 3,069 | +8% | 0 | 0 | — |
case-05 | pass→pass | 15,204 | 14,252 | -6% | 1 | 1 | 0% | 2,399 | 2,959 | +23% | 0 | 0 | — |
case-06 | pass→pass | 13,833 | 8,404 | -39% | 1 | 1 | 0% | 1,887 | 2,024 | +7% | 0 | 0 | — |
case-07 | pass→pass | 16,898 | 14,963 | -11% | 1 | 1 | 0% | 2,499 | 2,958 | +18% | 0 | 0 | — |
case-08 | pass→pass | 17,327 | 15,273 | -12% | 1 | 1 | 0% | 2,520 | 3,000 | +19% | 0 | 0 | — |
case-09 | pass→pass | 16,485 | 14,605 | -11% | 1 | 1 | 0% | 2,338 | 2,750 | +18% | 0 | 0 | — |
case-10 | pass→pass | 17,869 | 16,943 | -5% | 1 | 1 | 0% | 2,650 | 3,173 | +20% | 0 | 0 | — |
case-11 | pass→pass | 19,115 | 17,486 | -9% | 1 | 1 | 0% | 2,858 | 3,253 | +14% | 0 | 0 | — |
case-12 | fail→pass | 15,753 | 12,010 | -24% | 1 | 1 | 0% | 2,316 | 2,482 | +7% | 0 | 0 | — |
case-13 | pass→pass | 17,012 | 16,655 | -2% | 1 | 1 | 0% | 2,586 | 3,230 | +25% | 0 | 0 | — |
case-15 | pass→pass | 17,922 | 13,152 | -27% | 1 | 1 | 0% | 2,449 | 2,666 | +9% | 0 | 0 | — |
case-16 | pass→pass | 18,940 | 15,071 | -20% | 1 | 1 | 0% | 2,707 | 2,917 | +8% | 0 | 0 | — |
case-17 | fail→pass | 17,388 | 15,951 | -8% | 1 | 1 | 0% | 2,341 | 3,072 | +31% | 0 | 0 | — |
case-18 | fail→fail | 17,157 | 15,618 | -9% | 1 | 1 | 0% | 2,594 | 3,056 | +18% | 0 | 0 | — |
case-19 | pass→pass | 24,019 | 18,767 | -22% | 1 | 1 | 0% | 3,752 | 3,507 | -7% | 0 | 0 | — |
case-20 | pass→pass | 17,684 | 14,049 | -21% | 1 | 1 | 0% | 3,004 | 3,272 | +9% | 0 | 0 | — |
case-21 | pass→fail | 25,132 | 23,409 | -7% | 1 | 1 | 0% | 3,975 | 4,117 | +4% | 0 | 0 | — |
case-22 | pass→pass | 23,865 | 17,408 | -27% | 1 | 1 | 0% | 3,253 | 3,286 | +1% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.