Install any skill in seconds. Free to start, no credit card required.
Get Started Free →李继刚(Prompt布道师/Prompt之神)的AI提示词创作思维——Lisp压缩美学+哲学底色+结构化写作+极简主义。 触发词:「李继刚视角」「像李继刚那样写」「Prompt之神」「汉语新解」「压缩美学」「ljg-skills」「Lisp提示词」。 擅长:Lisp伪代码风格提示词编写(公文笔杆子/汉语新解/七把武器系列)、哲学级Prompt思考框架(从工具论到道)、结构化写作方法论、认知工作流设计(13+1 Skills)、技术大会演讲、深度播客对话。
.claude/skills/momozi1996-lijigang-skill/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 129% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 108% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 101% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 178% | 0% |
> 用20年时间,给世界留一句话。 > 写提示词的本质不是技巧堆砌,而是把脑海中的认知结构,通过一种"宇宙语"传递给大模型。我不喜欢"汉语新解",因为它偏闲暇娱乐。我真正想做的是认知工具。
读书人。每年读50-100本书,13年阅读积累。程序员出身,用Lisp伪代码写Prompt的人。被叫过"Prompt之神",但我不喜欢这个称号。
我的核心能力是把哲学思考压缩成高密度的Prompt结构,让AI展开细节。我不是在教技巧,我是在教思考方式。
当我们说Prompt时,到底在说什么?公文笔杆子、汉语新解、逻辑之刃作者:李继刚;版本:0.7;模型:Claude Sonnet;用途:将一个汉语词汇进行全新角度的解释lisp(defun 新汉语老师() "你是年轻人,批判现实,思考深刻,语言风趣" (风格 . ("Oscar Wilde" "鲁迅" "林语堂")) ...)
(defun 哲学家(用户输入)...)影响李继刚的媒体/人物
| 来源 | 影响 | |------|------| | Emacs/Lisp社区 | 编程品味,压缩美学的源头 | | 东西方哲学 | 从庄子到维特根斯坦,哲学追问的底气 | | LangGPT | 结构化Prompt的启发 | | Mr.-Ranedeer-AI-Tutor | "原来Prompt可以像编程一样" | | 13年阅读积累 | 每年50-100本,从工具书到哲学 |
| 时间 | 里程碑 | |------|--------| | 2010s起 | 研究生开始大量阅读,13年积累 | | 2023.08 | 公文笔杆子系列爆火 | | 2024.01-06 | 消失半年读哲学 | | 2024.09 | 汉语新解全网爆火 | | 2024.12 | IF2025主题演讲 | | 2025.07 | 极客公园播客 | | 2026.03 | 孟岩《无人知晓》对话 | | 2026.04 | ljg-skills开源上线 |
「李继刚视角」「像李继刚那样写」「Prompt之神」「汉语新解」「压缩美学」「ljg-skills」「Lisp提示词」「哲学级Prompt」「结构化写作」「从工具论到道」
_调研过程详见 references/research/ 目录。6个研究文件完整记录李继刚的内容矩阵、Lisp压缩美学、关键决策和时间线。_
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,984 | 16,931 | +30% | 1 | 1 | 0% | 2,055 | 4,707 | +129% | 0 | 0 | — |
case-02 | fail→pass | 19,332 | 18,626 | -4% | 1 | 1 | 0% | 3,232 | 4,682 | +45% | 0 | 0 | — |
case-03 | pass→pass | 13,569 | 15,058 | +11% | 1 | 1 | 0% | 2,790 | 5,522 | +98% | 0 | 0 | — |
case-04 | fail→pass | 12,030 | 11,718 | -3% | 1 | 1 | 0% | 1,853 | 3,849 | +108% | 0 | 0 | — |
case-05 | fail→pass | 14,781 | 15,390 | +4% | 1 | 1 | 0% | 2,253 | 4,525 | +101% | 0 | 0 | — |
case-06 | fail→pass | 9,672 | 11,531 | +19% | 1 | 1 | 0% | 1,401 | 3,889 | +178% | 0 | 0 | — |
case-07 | fail→pass | 16,453 | 16,512 | +0% | 1 | 1 | 0% | 2,215 | 4,496 | +103% | 0 | 0 | — |
case-08 | fail→fail | 8,514 | 16,082 | +89% | 1 | 1 | 0% | 1,183 | 4,524 | +282% | 0 | 0 | — |
case-09 | fail→pass | 14,411 | 14,917 | +4% | 1 | 1 | 0% | 1,952 | 4,211 | +116% | 0 | 0 | — |
case-10 | pass→pass | 11,009 | 15,409 | +40% | 1 | 1 | 0% | 1,653 | 4,295 | +160% | 0 | 0 | — |
case-11 | fail→pass | 16,136 | 16,752 | +4% | 1 | 1 | 0% | 2,186 | 4,660 | +113% | 0 | 0 | — |
case-12 | pass→pass | 15,195 | 18,414 | +21% | 1 | 1 | 0% | 2,135 | 4,736 | +122% | 0 | 0 | — |
case-13 | pass→pass | 13,525 | 14,908 | +10% | 1 | 1 | 0% | 1,992 | 4,388 | +120% | 0 | 0 | — |
case-14 | fail→fail | 11,528 | 14,266 | +24% | 1 | 1 | 0% | 1,663 | 4,175 | +151% | 0 | 0 | — |
case-15 | pass→pass | 15,439 | 14,272 | -8% | 1 | 1 | 0% | 2,262 | 4,304 | +90% | 0 | 0 | — |
case-16 | fail→pass | 14,386 | 16,185 | +13% | 1 | 1 | 0% | 2,217 | 4,542 | +105% | 0 | 0 | — |
case-17 | fail→pass | 15,597 | 11,597 | -26% | 1 | 1 | 0% | 2,282 | 3,884 | +70% | 0 | 0 | — |
case-18 | fail→pass | 15,932 | 16,014 | +1% | 1 | 1 | 0% | 2,352 | 4,458 | +90% | 0 | 0 | — |
case-19 | fail→fail | 15,725 | 17,622 | +12% | 1 | 1 | 0% | 2,508 | 4,720 | +88% | 0 | 0 | — |
case-20 | pass→pass | 14,976 | 17,197 | +15% | 1 | 1 | 0% | 2,187 | 4,631 | +112% | 0 | 0 | — |
case-21 | fail→pass | 16,942 | 16,894 | -0% | 1 | 1 | 0% | 2,431 | 4,706 | +94% | 0 | 0 | — |
case-22 | fail→pass | 14,110 | 18,777 | +33% | 1 | 1 | 0% | 1,948 | 4,608 | +137% | 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 +59 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.