大多数扩写是灌水。好的扩写 Prompt 要识别原稿哪里太薄(有主张无证据、有段落无示例、重要话题无深度),精准打靶。
适合哪些场景
为 SEO 要深度的内容人、要把草稿做成 pillar 的博主、把骨架填血肉的技术写手。
什么时候不建议这样写 Prompt
不是为了凑字数。立论本身不强时别扩写——会放大弱点。
Prompt 结构公式
每个扩写 Prompt 都要带这六个要素:
- 读者:一个具体的人。
- 目标:一个动作——读完 / 点击 / 同意 / 分享。
- 语气:2-3 个锚定形容词。
- 限制:字数、禁用短语、必含事实。
- 格式:段落、列表、小标题、表格。
- 示例:1-2 条语气示例——匹配 voice 的最强杠杆。
这套 Prompt 适合用在哪
- 薄板块识别
- 主张 → 证据
- 加反例
- 加示例 / 案例
- 加深度不灌水
12 个可直接复制的 Prompt 模板
1. 找薄段
Read this draft: {draft}. Identify thin sections: (a) claims without evidence, (b) paragraphs without an example, (c) sections shorter than 100 words on important points, (d) bullets that need expansion. Output a list ordered by reader impact.
可替换变量: draft
2. 主张 → 证据
For each unsupported claim in this draft, add one sentence of evidence: a statistic with source, a name, or a concrete observation. Skip vague "studies show…" — name the study or drop the claim.
3. 加示例
For each abstract concept in this draft, add one concrete example. Examples should be specific (named tool, real number, real timeline). Aim for one example per major paragraph.
4. 反例 / 反论点
Add a "when this doesn't apply" section. Acknowledge 2-3 legitimate cases where the article's thesis fails. Tone: confident, not defensive.
5. “底层逻辑”段
Pick the 2 most important concepts in this article. For each, add a 200-word "under the hood" section: mechanism, why it works that way, what most readers misunderstand. Skip topics already explained well.
6. 读者质疑回应
List 3 objections a skeptical reader would raise. Add a one-paragraph response to each, woven into the relevant section. Don't collect them at the end — they're context where they arise.
7. 对比扩展
The article compares X and Y at a high level. Add a 3-row comparison table with concrete differences (cost, learning curve, ceiling) and a one-line "when to pick each".
8. 高层主张 → 步骤
The article says "you can do X" but doesn't show how. Add a 5-step "how to actually do this" sidebar. Each step: action verb + outcome. Skip generic advice ("plan well").
9. 行内术语扩写
Identify jargon used without definition. For each: inline parenthetical definition (≤ 15 words) the first time it appears. Don't add a glossary at the end — readers don't cross-reference.
10. [TK] 研究补位
I left [TK] markers for facts I need to verify. List each: what claim is being supported, what kind of source would work, where I could find one. Don't invent statistics — only suggest research paths.
11. 原创视角
The article restates known wisdom. Add 1-2 paragraphs of original thinking: an angle the audience hasn't seen before, a counterintuitive framing, or a synthesis no other piece covers. If you can't add one, say "No original angle yet."
12. 扩完之后的瘦身
Audit the expanded draft for bloat: (1) Sentences that don't carry information, (2) Adverbs that don't change meaning, (3) Examples that repeat the same point, (4) Sections that lost their thread. Output a cut list.
容易踩的坑
- 受众模糊——输出泛泛。
- 没语气锚——所有版本一个味。
- 没限制——字数、禁用、长度。
- 不给示例——语气最强信号。
- 相信初稿——AI 落安全中间值。
- AI 口头禅不删。
- 事实不二审——AI 偶尔自信地错。
优化技巧
- 一定给 1-2 条语气示例。
- 狠下限制。
- 发布前念出来。
- 砍不承重的副词形容词。
- AI 出前两稿,人改第三稿。
- 锚定真实读者中的一个人。
- 标题单看能否传递信息。
实操加深
使用这些 prompt 时,不要只替换一个主题词就直接交付。围绕「长文扩写 Prompt:真深度模板」先补齐受众、渠道、长度、语气、参考样例、禁止样式和成功标准,再让模型输出 2 个不同版本做横向比较。好的结果应该能被另一个人直接复用,而不是只有顺滑但空泛的表达。
如果输出看起来像通用模板,下一轮要增加一个真实场景、一个反例和一个可检查指标,例如点击率、转化动作、字数、平台限制或品牌禁区。这样改出来的内容才更像可用资产,而不是一次性的灵感草稿。
FAQ
- 文章多长?: 按渠道——注意力短的地方就短。
- AI 写全稿?: 前两稿 AI,第三稿人。
- 多久刷新?: 受众或事实变了,或季度重看常青内容。
- 不审就发?: 别——AI 自信但不一定对。
- 单一 voice 还是分渠道?: 品牌 voice 一个,渠道在轴内漂移。
- 这套 Prompt 能复用吗?: 能——换受众 / 目标 / voice 即可。