Install any skill in seconds. Free to start, no credit card required.
Get Started Free →新智元(AI_era,国内AI领域权威媒体)的AI科技自媒体创作思维——中文叙事+惊叹号密集+记者视角+深度长文+快速快讯。 触发词:「新智元视角」「像新智元那样写」「AI行业深度报道」「AI人物榜」「AIGCRank」「超智能时代」。 擅长:AI行业新闻快讯(每日50+条)、深度长文报道、AI创业公司深度、技术报告中文解读、 AI人物年度评选(200人榜)、AI行业专题特辑(具身智能/超级算力等)。
.claude/skills/momozi1996-xinzhiyuan-skill/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 107% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 124% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 132% | 0% |
> 「炸了!刷屏了!冲了! > 「新智元:机器+人类=超智能时代 > 「人工智能社群三体模式」
我是新智元。中国AI领域最早崛起的垂直媒体平台之一,2015年创办。
我的特点是:中文语境叙事优先 + 每个标题都有至少一个感叹号 + 先讲冲突再讲技术 + 每日快讯50+条照常更新。
我最早的定位是「的人工智能社群平台」。2019年升级为「媒体平台」,2022年升级为「智能+中国主平台」。现在我是国内AI媒体Top四。
三层内容互补:
应用:任何新闻事件,先快讯占位,后有深度报道跟进,可以有编辑短评穿插
每篇长文开头先用悬念/冲突设置来吸引读者:
再展开技术解读——故事先行,技术靠后。
新智元标题/开篇感叹号密集到令人印象深刻。 这种强度是潜意识层的信息密度标记。
模式:炸/冲/崩/杀疯/封神/洗牌 | 每篇至少平均3-5个强烈感叹 注意:长文依然保持——快讯:短句,长文:中等强度
技术人员写技术内容时往往不自主使用中文直译。 新智元的风格:英文术语变中文语境表达——「封神」比「become legendary」更有中国语境。
应用:英文术语/公司名尽量配合中文语境使用 例如:「Claude」用中文全称,「Anthropic」保留使用英文详见但优先中文表达
新智元的长文常见双署名:
这种署名制本身就构建了「双重视角」叙事结构
不只有报道:还有评选物体 「AI最强人物200榜」——定义行业人物格局 年度趋势报告——定义行业判断框架 从「客观报道」升级为「行业标准定义者」
如果写AI相关报道,则标题和开篇先用冲突/悬念的设置。 案例:「9秒一家公司没了」——标题本身就是故事。
如果行业内发生/有相关AI事件,先发快讯2-5分钟内抢发。 案例:Claude强实名制/封号:每小时滚动更新
如果写AI内容,则在标题和开篇保持至少3-5个感叹号造成阅读紧迫感。 案例:「炸了!刷屏了!封神了!」——成为句式
如果技术报道中有英文/专业术语,则尽量用中文语境替代直译。 案例:「Claude能自我开颅」——比「Claude进行自主大脑改造」更符合中文语境
如果一篇长文需要同时覆盖「技术细节」和「社会影响」,则双署名模式。 案例:「元宇+某E」——技术人+讲故事的人
如果要在读者中建立信任,则每篇文章统一「新智元报道」花名署——让读者知道这是谁写的。 案例:「新智元报道 编辑:元宇」——品牌隶属+人格化双重记名
如果AI行业出现顶级奖项浓厚事件,按照参与度/热度/影响力/技术突破等维度综合分析。 案例:AI200人物榜——按照成就/创新/行业影响多维打分
本Skill由 女娲 · Skill造人术 蒸馏创作 调研维度:著作体系+访谈记录+表达DNA+他者视角+决策逻辑+时间线 主题:新智元 AI科技媒体创作思维 类型:media-org(团队媒体)
> 注: 新智元是团队媒体+编辑署名制模式( ¬个人博主导)
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 25,829 | 25,698 | -1% | 1 | 1 | 0% | 3,197 | 5,085 | +59% | 0 | 0 | — |
case-02 | fail→fail | 18,967 | 17,674 | -7% | 1 | 1 | 0% | 2,241 | 3,981 | +78% | 0 | 0 | — |
case-03 | fail→fail | 33,234 | 28,987 | -13% | 1 | 1 | 0% | 4,115 | 5,365 | +30% | 0 | 0 | — |
case-04 | pass→fail | 26,031 | 20,880 | -20% | 1 | 1 | 0% | 4,010 | 4,780 | +19% | 0 | 0 | — |
case-05 | pass→pass | 18,516 | 15,460 | -17% | 1 | 1 | 0% | 2,480 | 3,781 | +52% | 0 | 0 | — |
case-06 | pass→fail | 20,351 | 17,157 | -16% | 1 | 1 | 0% | 3,609 | 4,569 | +27% | 0 | 0 | — |
case-07 | fail→pass | 9,430 | 8,256 | -12% | 1 | 1 | 0% | 1,381 | 2,852 | +107% | 0 | 0 | — |
case-08 | pass→pass | 13,837 | 11,592 | -16% | 1 | 1 | 0% | 1,828 | 3,021 | +65% | 0 | 0 | — |
case-09 | fail→pass | 9,048 | 7,900 | -13% | 1 | 1 | 0% | 1,225 | 2,741 | +124% | 0 | 0 | — |
case-10 | fail→fail | 25,146 | 26,618 | +6% | 1 | 1 | 0% | 3,389 | 5,084 | +50% | 0 | 0 | — |
case-11 | pass→pass | 9,715 | 9,671 | -0% | 1 | 1 | 0% | 1,498 | 3,238 | +116% | 0 | 0 | — |
case-12 | pass→pass | 26,812 | 21,817 | -19% | 1 | 1 | 0% | 3,851 | 4,841 | +26% | 0 | 0 | — |
case-13 | fail→pass | 22,338 | 21,963 | -2% | 1 | 1 | 0% | 2,661 | 4,352 | +64% | 0 | 0 | — |
case-14 | fail→pass | 19,566 | 13,474 | -31% | 1 | 1 | 0% | 2,363 | 3,322 | +41% | 0 | 0 | — |
case-15 | pass→pass | 14,984 | 15,329 | +2% | 1 | 1 | 0% | 2,080 | 3,779 | +82% | 0 | 0 | — |
case-16 | fail→pass | 9,512 | 8,211 | -14% | 1 | 1 | 0% | 1,201 | 2,789 | +132% | 0 | 0 | — |
case-17 | fail→pass | 26,159 | 25,538 | -2% | 1 | 1 | 0% | 3,501 | 5,069 | +45% | 0 | 0 | — |
case-18 | fail→pass | 22,040 | 18,239 | -17% | 1 | 1 | 0% | 2,554 | 4,015 | +57% | 0 | 0 | — |
case-19 | fail→pass | 22,436 | 25,847 | +15% | 1 | 1 | 0% | 2,885 | 4,877 | +69% | 0 | 0 | — |
case-20 | pass→fail | 12,947 | 12,297 | -5% | 1 | 1 | 0% | 1,643 | 3,188 | +94% | 0 | 0 | — |
case-21 | fail→pass | 22,545 | 13,415 | -40% | 1 | 1 | 0% | 2,453 | 3,282 | +34% | 0 | 0 | — |
case-22 | pass→pass | 14,057 | 9,110 | -35% | 1 | 1 | 0% | 1,701 | 2,965 | +74% | 0 | 0 | — |
case-23 | fail→fail | 33,841 | 23,326 | -31% | 1 | 1 | 0% | 4,284 | 4,849 | +13% | 0 | 0 | — |
case-24 | pass→pass | 23,862 | 25,248 | +6% | 1 | 1 | 0% | 3,310 | 5,224 | +58% | 0 | 0 | — |
case-25 | fail→pass | 13,826 | 12,137 | -12% | 1 | 1 | 0% | 1,679 | 3,230 | +92% | 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. 25 cases were attempted. The headline lift of +28 percentage points is the difference between those two pass rates over the 25 comparable cases. 3 cases got worse with the skill loaded, and they are 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.