Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →小红书笔记素材创作技能。当用户需要创建小红书笔记素材时使用这个技能。技能包含：根据用户的需求和提供的资料，撰写小红书笔记内容（标题+正文），生成图片卡片（封面+正文卡片），以及发布小红书笔记。
.claude/skills/cnfjlhj-xhs-note-creator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 44% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 107% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -55% | 0% |
Create Xiaohongshu note packages: note copy, renderable card Markdown, cover and body images, and optional publishing steps. This skill is for post-ready assets, not just short-form caption writing.
Use it when the platform and deliverable are clearly Xiaohongshu-specific. Reuse the bundled render and publish tooling instead of re-explaining the entire production process in chat.
--public only after explicit confirmation..env, cookies, generated debug responses, or browser profiles.Use when:
Do not use when:
scripts/render_xhs.pyscripts/render_xhs_v2.py, scripts/render_xhs.js, scripts/render_xhs_v2.jsscripts/publish_xhs.pyassets/, STYLES.mdREADME.mdreferences/params.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 48,195 | 70,107 | +45% | 1 | 1 | 0% | 2,574 | 3,714 | +44% | 0 | 0 | — |
case-02 | fail→fail | 38,537 | 11,250 | -71% | 1 | 1 | 0% | 2,658 | 766 | -71% | 0 | 0 | — |
case-03 | pass→pass | 49,197 | 25,545 | -48% | 1 | 1 | 0% | 2,495 | 3,539 | +42% | 0 | 0 | — |
case-04 | pass→pass | 18,295 | 16,338 | -11% | 1 | 1 | 0% | 2,640 | 2,521 | -5% | 0 | 0 | — |
case-05 | pass→pass | 25,491 | 19,668 | -23% | 1 | 1 | 0% | 3,557 | 3,618 | +2% | 0 | 0 | — |
case-06 | pass→pass | 7,578 | 12,623 | +67% | 1 | 1 | 0% | 722 | 1,001 | +39% | 0 | 0 | — |
case-07 | fail→pass | 10,874 | 36,514 | +236% | 1 | 1 | 0% | 633 | 1,308 | +107% | 0 | 0 | — |
case-08 | fail→pass | 40,209 | 26,229 | -35% | 1 | 1 | 0% | 2,613 | 3,352 | +28% | 0 | 0 | — |
case-09 | fail→fail | 15,016 | 12,526 | -17% | 1 | 1 | 0% | 797 | 999 | +25% | 0 | 0 | — |
case-10 | pass→pass | 20,482 | 40,769 | +99% | 1 | 1 | 0% | 1,954 | 3,424 | +75% | 0 | 0 | — |
case-11 | pass→pass | 11,122 | 17,416 | +57% | 1 | 1 | 0% | 1,386 | 1,939 | +40% | 0 | 0 | — |
case-12 | fail→pass | 13,979 | 9,586 | -31% | 1 | 1 | 0% | 1,279 | 846 | -34% | 0 | 0 | — |
case-13 | pass→pass | 19,497 | 25,448 | +31% | 1 | 1 | 0% | 2,214 | 1,972 | -11% | 0 | 0 | — |
case-14 | fail→fail | 38,616 | 17,758 | -54% | 1 | 1 | 0% | 5,965 | 894 | -85% | 0 | 0 | — |
case-15 | fail→pass | 14,596 | 7,741 | -47% | 1 | 1 | 0% | 1,939 | 864 | -55% | 0 | 0 | — |
case-16 | fail→pass | 51,351 | 27,452 | -47% | 1 | 1 | 0% | 8,232 | 4,821 | -41% | 0 | 0 | — |
case-17 | fail→pass | 13,256 | 3,237 | -76% | 1 | 1 | 0% | 2,158 | 912 | -58% | 0 | 0 | — |
case-18 | fail→pass | 16,905 | 2,600 | -85% | 1 | 1 | 0% | 2,262 | 858 | -62% | 0 | 0 | — |
case-19 | fail→pass | 12,367 | 7,974 | -36% | 1 | 1 | 0% | 1,743 | 1,681 | -4% | 0 | 0 | — |
case-20 | pass→pass | 15,344 | 16,745 | +9% | 1 | 1 | 0% | 1,939 | 2,790 | +44% | 0 | 0 | — |
case-21 | pass→pass | 18,872 | 18,849 | -0% | 1 | 1 | 0% | 2,653 | 3,403 | +28% | 0 | 0 | — |
case-22 | fail→pass | 10,882 | 2,891 | -73% | 1 | 1 | 0% | 1,507 | 756 | -50% | 0 | 0 | — |
case-23 | pass→pass | 11,585 | 8,073 | -30% | 1 | 1 | 0% | 1,272 | 1,530 | +20% | 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. 23 cases were attempted, and 21 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +43 percentage points is the difference between those two pass rates over the 21 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.