netease-youdao/article-writer

Multi-Style Article Creation

Use Cases

• User says "写一篇关于XX的文章"
• User says "帮我写一篇公众号文章"
• User selects a topic from the content calendar to start writing
• User specifies a writing style (e.g., "深度分析风格", "写个教程")

5 Writing Styles

| Style ID | Name | Characteristics | Word Count | Use Cases | |----------|------|-----------------|------------|-----------| | deep-analysis | 深度分析 | Rigorous structure, data-backed | 2000-4000 words | Trend analysis, in-depth reporting | | practical-guide | 实用指南 | Clear steps, highly actionable | 1500-3000 words | Tool tutorials, how-to guides | | story-driven | 故事驱动 | Conversational, emotional resonance | 1500-2500 words | Personal stories, case reviews | | opinion | 观点评论 | Sharp opening, pros/cons argumentation | 1000-2000 words | Hot takes, controversial topics | | news-brief | 新闻简报 | Inverted pyramid, fact-focused | 500-1000 words | Breaking news, information roundups |

Workflow

Step 1: Read the Brief

Obtain topic information from:

1. Entries with status planned in content_calendar.json
2. Topic description directly provided by the user

Extract key information:

• Topic direction / title
• Target audience
• Writing style (if not specified, recommend based on topic content)
• Reference material URLs

Step 2: Determine Writing Style

If user hasn't specified, recommend based on topic:

| Topic Characteristics | Recommended Style | |----------------------|-------------------| | Involves data, trends, underlying causes | deep-analysis | | "How to", "tutorial", "steps" | practical-guide | | Involves people, experiences, insights | story-driven | | Involves controversy, hot topic commentary | opinion | | Breaking events, quick information | news-brief |

Confirm the style choice with the user.

Step 3: Material Collection

Use content-planner's search script to collect reference materials:

node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "topic keywords" -n 10

Also use web-search skill for additional materials.

Organize material list:

• Citable data/statistics
• Reference cases/stories
• Facts that need verification

Step 4: Generate Outline

Generate an outline based on the selected style using the corresponding structure template.

deep-analysis Template

## 引入 (200-300字) — 反直觉数据/现象开头
## 背景 (300-500字) — 事件/现象的来龙去脉
## 分析维度1 (400-600字) — 核心论点 + 数据支撑
## 分析维度2 (400-600字) — 对比/反面论证
## 分析维度3 (400-600字) — 深层原因 + 影响预测
## 总结与展望 (200-300字) — 核心观点 + CTA

practical-guide Template

## 开头 (100-200字) — 痛点共鸣 + 承诺
## 前置准备 (200-300字)
## 步骤1 (300-500字) — 具体操作 + 常见坑
## 步骤2 (300-500字) — 具体操作 + 关键参数
## 步骤3 (300-500字) — 具体操作 + 验证方法
## 进阶技巧 (200-300字)
## 总结 (100-200字) — FAQ + CTA

story-driven Template

## 开头 (150-200字) — 场景切入 + 悬念
## 背景铺垫 (200-300字) — 人物/背景/冲突
## 转折1 (300-400字) — 关键事件 + 感受
## 转折2 (300-400字) — 新尝试 + 结果
## 高潮 (200-300字) — 核心洞察
## 结尾 (150-200字) — 启发 + CTA

opinion Template

## 锐利开头 (100-150字) — 直接亮观点
## 现象描述 (200-300字) — 主流观点
## 正面论证 (300-400字) — 我的论点 + 数据
## 反面回应 (200-300字) — 预设反驳 + 反驳
## 深度思考 (200-300字) — 本质 + 影响
## 总结 (100-150字) — 重申观点 + CTA

news-brief Template

## 核心信息 (100-200字) — What/When/Where/Who
## 事件详情 (200-300字)
## 背景 (100-200字)
## 反应 (100-200字)
## 编者按 (50-100字)

Step 5: User Approval of Outline

This is a mandatory approval gate and cannot be skipped.

Present the outline to the user and ask for confirmation or modification requests.

Step 6: Write the Content

After approval, write content paragraph by paragraph following the outline.

Universal Writing Rules:

1. Use stories instead of preaching

   • ❌ "风险管理很重要，应该做应急预案"
   • ✅ "去年，我的创业团队差点因为一个核心员工离职而崩溃，因为我们没有任何备选方案。"

2. Use analogies and metaphors

   • ❌ "分布式系统很复杂"
   • ✅ "分布式系统就像连锁餐厅——每个分店需要协作，同时又要独立运营。"

3. Support with data but don't pile it on

   • ❌ "据IDC报告，全球AI市场2023年增长45%，预计2025年达1000亿美元..."
   • ✅ "AI市场疯狂增长——每年翻一番。但在增长背后，真正赚钱的公司不到5家。"

4. State opinions directly, avoid ambiguity

   • ❌ "有人认为...也有人认为...各有道理"
   • ✅ "说实话，我认为XX的做法是错的，因为..."

5. Use short sentences and line breaks

Data integrity rules:

• If specific data is needed but uncertain, mark [数据待确认] and confirm with user
• Use search tools to verify key facts
• Can tell fictional stories using "我见过..." or "一个朋友...", but don't fabricate data

Step 7: Formatting Optimization

WeChat Formatting Hard Rules:

1. Paragraphs no more than 4 lines (mobile screen visible range)
2. Insert a subheading or bold sentence every 3-4 paragraphs
3. Must have a hook within the first 3 lines (question, data, story, counter-intuitive viewpoint)
4. Must have a clear CTA at the end (follow/share/comment prompt)

Markdown Formatting Standards:

• No first-line indentation, use blank lines to separate paragraphs
• Maximum 2 heading levels (##), no deep nesting
• Bold only the 1-2 most important words per paragraph
• Use quotes for data, golden sentences, or important viewpoints
• Lists maximum 5 items

Step 8: Output Draft

Save the article as Markdown:

Filename format: drafts/YYYYMMDD_[topic-slug].md

---
title: Article Title
date: YYYY-MM-DD
style: deep-analysis
summary: Article summary (within 100 words)
---

## Opening

Content...

Step 9: Quality Checklist

✅ Title — Sparks curiosity or resonance
✅ Opening — First 100 words are engaging
✅ Body — Has 2-3 clear viewpoints
✅ Cases — Uses stories not preaching
✅ Formatting — Easy to read (short paragraphs, bold, subheadings)
✅ Word Count — Within style-specified range
✅ CTA — Ending has action prompt

Title Optimization Process

Step 1: Generate 10 Candidate Titles

Use these psychological strategies:

| Strategy | Description | Example | |----------|-------------|---------| | Suspense | Spark curiosity | "为什么我放弃了年薪50万的工作" | | Benefit | Clarify reader gains | "掌握这3个技巧，效率提升200%" | | Pain Point | Hit reader anxiety | "别让这个习惯毁了你的职业生涯" | | Numbers | Specific and tangible | "50%的人都误解了这个真相" | | Rhetorical | Stimulate thinking | "你真的了解AI吗？" | | Contrast | Create contrast | "BAT vs 创业公司：差别在哪" |

Step 2: Score and Filter

Score on 3 dimensions (Attractiveness 40%, Shareability 30%, SEO 30%) and present top 3 to user.

Integration with Other Skills

• Upstream: content-planner's content_calendar.json provides topic input
• Downstream: Markdown files in drafts/ can be further processed