Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› sci-writing
sci-writing
GitHub指导AAAS《科学》期刊写作，区分Report与研究文章格式，控制字数与图表上限。强调将方法移至补充材料，按论证逻辑而非时间顺序组织结果，并精简语言以符合期刊严格的篇幅要求。
触发场景
不确定作品是Report还是Research Article
正文超字数或方法部分过于冗长
结构混乱，未按论证逻辑组织
论文篇幅超出预算限制
安装
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill sci-writing -g -y
SKILL.md
Frontmatter
{
"name": "sci-writing",
"description": "Use when choosing the Science (AAAS) format (Research Article vs Report) and holding its length and structure budgets — main-text word caps, figure caps, argument-ordered Results, and deciding what belongs in the main text versus Supplementary Materials."
}
Main-Text Writing & Format (sci-writing)
When to trigger
- Unsure whether the work is a Report or a Research Article.
- The main text is over length, or Methods are bloating the body.
- The structure wanders (lab-notebook order instead of argument order).
- Co-authors keep adding paragraphs and the paper is creeping past budget.
Pick the format first
| Format | Use when | Main-text target | Display items |
|---|---|---|---|
| Report | A single, decisive advance told quickly | ~2,500 words | ≤ 4 (figs+tables) |
| Research Article | A larger study needing more development/multiple results | ~4,500 words | ≤ 6 (figs+tables) |
Numbers are working targets; confirm current caps in the author guidelines. The principle is fixed: Science main texts are short, and most detail lives in Supplementary Materials.
What goes where
- Main text (Report): no section headers beyond the flow; combined results-and-discussion narrative. Methods are NOT in the main text — they go to Supplementary Materials (Materials and Methods).
- Main text (Research Article): may use brief headers; still lean. Extended methods → Supplementary.
- Supplementary Materials: full Materials and Methods, supplementary figures/tables, supplementary text, captions, and additional data. This is where rigor lives without bloating the body.
Structure as argument, not chronology
Order the Results by the logic of the claim, not the order experiments were run:
- Establish the phenomenon (the headline result).
- Rule out the obvious alternative explanations.
- Show the mechanism / generality.
- Demonstrate the broad implication.
Each Results paragraph: claim sentence first, then the evidence (figure callout + numbers), then the inference.
Length discipline tactics
- One idea per paragraph; lead sentence is the claim.
- Move any sentence that a general reader can skip → Supplementary.
- Methods detail, validation, controls → Supplementary (cited as "fig. S3", "table S1").
- Cut "In order to" → "To"; cut "It is worth noting that"; cut throat-clearing.
Before/after: compressing to the Science register
Science prose earns its word budget sentence by sentence. Typical compressions:
| Draft (lab-report register) | Science register |
|---|---|
| "In order to investigate whether X affects Y, we performed a series of experiments in which…" | "To test whether X affects Y, we…" |
| "It is worth noting that the observed effect was statistically significant (p < 0.05)." | "The effect was robust (Δ = 12 ± 2%, P = 0.003, n = 14)." |
| "As can be seen in Fig. 2, there appears to be a trend toward increased activity." | "Activity increased threefold (Fig. 2)." |
| "These results may suggest a possible role for Z in this process." | "Z is required for this process (fig. S4)." |
The pattern: verbs over nominalizations, numbers over adjectives, one hedge maximum per claim, and every figure callout attached to a quantitative statement rather than "as shown in".
What Science referees expect from the body text
- Each Results paragraph is checkable: a referee can match the claim sentence to a specific panel and a specific statistic without hunting.
- Alternative explanations are ruled out in the main text, even if the ruling-out data live in the SM ("this was not due to A or B; figs. S5 and S6").
- The final paragraph states scope and limits in one or two sentences — referees at a general-science weekly flag manuscripts whose conclusions quietly exceed the shown systems or conditions.
- No forward references to unpublished work as load-bearing evidence.
Reference and supplement cross-refs
- Main figures: "Fig. 1"; supplementary: "fig. S1" (lowercase f); tables: "Table 1" / "table S1".
- Supplementary Materials are cited in the main text and listed at the end.
Output format
【Format】 Report (~2,500w / ≤4 items) or Research Article (~4,500w / ≤6 items)
【Current main-text word count】 N → over/under budget by M
【Methods location】 main text (FIX → move to SM) / Supplementary (ok)
【Results order】 argument-ordered? yes/no
【Items in main vs SM】 main: [...] / SM: [...]
【Next】 sci-figures
Anti-patterns
- Do not keep a full Methods section in a Report's main text.
- Do not structure Results chronologically when a logical order is clearer.
- Do not treat the word cap as negotiable by shrinking font in figures — move content to SM instead.
- Do not add a standalone Discussion that repeats Results; integrate the interpretation.
版本历史
- 1839142 当前 2026-07-05 14:24
