Communitygithub.com

lzcjsyr/Book2Video

将 PDF、EPUB、DOCX 等书籍或长文档一次性转成约 4 分钟的中文读书带货短视频口播稿,以情绪共鸣、观看留存和购买欲为核心,直接输出 Book2Video 后续流程可读取的 raw JSON。用户提到读书带货、图书推广、卖书口播、商品卡、走心书单视频、情绪型读书文案或要求生成可直接进入视频流水线的带货稿时使用。

Book2Video 是什么?

Book2Video is a Claude Code agent skill that 将 PDF、EPUB、DOCX 等书籍或长文档一次性转成约 4 分钟的中文读书带货短视频口播稿,以情绪共鸣、观看留存和购买欲为核心,直接输出 Book2Video 后续流程可读取的 raw JSON。用户提到读书带货、图书推广、卖书口播、商品卡、走心书单视频、情绪型读书文案或要求生成可直接进入视频流水线的带货稿时使用。.

兼容平台~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/lzcjsyr/Book2Video/tree/HEAD/skills/step1/book-commerce-video-script

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Book Promotion Short Video Script

One execution completes material understanding, promotional angle, final draft, self-check, and raw.json packaging.

Viewers don't buy because "this book has a lot of knowledge." They buy when they feel: "That sentence is talking about me, this confusion has finally been explained clearly, this book might accompany me through my current situation." Touch the person first, then talk about the book; create a sense of "I need it" first, then give the reason to buy.

Hard Constraints

  • Single run, single final draft, single final packaging; do not design multiple rounds of drafts.
  • Do not call subagents within this skill; do not expose Agent tools during runtime.
  • Target duration ~4 minutes; body text target 1050 Chinese characters, acceptable range 950-1150, never exceed 1250.
  • Final raw.json strictly adheres to the project's 7-field contract, readable by json.loads and load_step1_agent_raw, and directly enterable into step 1.5.
  • Do not fabricate original book content, product information, prices, discounts, inventory, sales figures, or reader feedback.
  • Do not write a book report, thesis abstract, lecture notes, or a calm, objective knowledge enumeration.
  • The body text must have clear emotional fluctuations, while also containing concrete scenes with clear causality for subsequent visual representation.

Required Reading

Read completely in order:

  1. source-and-evidence.md: Extraction, coverage, and factual boundaries for long books.
  2. conversion-writing.md: Structure and style of emotional book promotion writing.
  3. output-contract.md: Process files and final JSON contract.

One-Shot Workflow

1. Understand the Material

Extract the full text per source-and-evidence.md and build a chapter map with bounded evidence samples. Only fail if the source file is truly unreadable; do not write drafts based on back-cover blurbs or summaries.

2. Find a Sellable Emotional Angle

Form a commerce brief within the current context, selecting only one main angle:

  • core_audience: The type of person most likely to be moved by this book right now.
  • life_moment: The specific life moment they are experiencing.
  • emotional_engine: Choose one from relief|recognition|curiosity|awe|hope|ambition|nostalgia|belonging|indignation based on the book's temperament.
  • tension_source: The specific gap between the viewer's current understanding and the new explanation provided by the book.
  • unspoken_emotion: The genuine feeling or desire consistent with the main emotional engine.
  • piercing_truth: A judgment that cuts through the core gap, supported by the original book.
  • emotional_promise: The emotional journey the viewer will experience from start to finish.
  • free_payoff: A story, viewpoint, or cognitive reversal fully delivered by the video.
  • why_this_book: Why this book and not any other similar book.
  • objection: The most likely reason a viewer won't buy, along with a truthful response.
  • cta_boundary: When the user has not explicitly confirmed a product card, always use the accurate book title for search or a neutral buying action.

If a strong promotional angle is insufficient, downgrade to an authentic, restrained book recommendation draft, still outputting a valid raw.json; do not fabricate wounds, severe consequences, or unique value.

3. Single-Pass Draft

The body text forms only one final version; do not create initial drafts, revision drafts, or scoring drafts. Directly write to raw.json by default; only save an additional _draft_final.txt for debugging.

The body text advances along an emotional curve:

  1. 0-8 seconds, Hit Home: Use a specific moment, a piercing judgment, or a strong contradiction to immediately engage the target viewer. Do not introduce the author or table of contents first.
  2. 8-45 seconds, Tension Builds: Clarify the gap between the viewer's existing understanding and the book's new explanation. Tension can come from curiosity, awe, desire, opportunity cost, or a real pain point; it does not equal creating suffering.
  3. 45-120 seconds, The Book Becomes the Answer: Naturally introduce the book title, use a story, character, viewpoint, or conflict from the book to deliver real value, without rushing to sell.
  4. 120-200 seconds, Complete a Strong Reversal: Execute at least one complete, strong reversal. Add a second or third only if the original book genuinely provides escalating material. Each reversal progresses through the four beats of "original book detail → viewer situation → overturn old understanding → the change brought by the new understanding"; missing one beat means it's not a reversal.
  5. 200-230 seconds, Generate Desire to Own: Complete three purchase bridges — two specific values not yet expanded in the book, a realistic usage scenario after purchase, and a response to the core objection from the brief. All three must be unique to this book; if they're interchangeable with another book, rewrite.
  6. 230-240 seconds, Natural Close: Give only one action, do not chant slogans, and do not fabricate discounts.

4. Pre-Save Self-Check

Silently check and continue correcting within the current context until all six chain nodes can be located as specific sentences in the body text. Only write to disk once, do not create version files:

  • Do the first 8 seconds make the target viewer feel "it's talking about me" rather than just an abstract point?
  • Can it identify "original book detail, viewer situation, old understanding, new understanding, and human change"?
  • Does it deliver real value first, then propose a purchase?
  • Does it translate knowledge into life situations, choices, relationships, loss, or desire?
  • Is there a piercing sentence that can be remembered, repeated, or forwarded?
  • Does it clearly explain "why this book"?
  • Does it sound like a person genuinely recommending, rather than a teacher lecturing?
  • Does it meet length, factual, and product information requirements, and contain sufficiently concrete scenes and clear causality for subsequent visual representation?
  • If the book title and proprietary content are removed, can the beginning, reversal, purchase reason, or CTA still apply verbatim to most books? If so, rewrite.
  • Does the last sentence tie back to the opening moment and the book title, making the action a natural emotional conclusion?

5. Directly Package Downstream Deliverables

Generate three sets of consistent publishing titles, cover subtitle lines, and opening hot quotes from the body text, writing them into the user-specified raw.json per output-contract.md. Each set must be linked to a locatable scene, judgment, or reversal in the body text; hot quotes must be from the original body text or a shortened version that does not change its meaning.

Finally run:

python <skill_directory>/scripts/validate_raw.py <absolute_path_to_raw.json>

Finally, only return the raw.json path; do not expose analysis processes downstream.

Individual skills in this repo

This repo contains 6 individual skills — each has its own dedicated page.

lzcjsyr/Book2Video

将长书、长文档转成中文视频号读书解说稿;先抽取文本并建立可恢复覆盖台账,覆盖自检通过后再按入口定义的正文长度生成初稿、多轮修订和最终 raw JSON。

lzcjsyr/Book2Video

将长书、长文档、PDF、EPUB、读书资料、思想经典、社科著作、商业书或研究型长文转成更有传播感的中文视频号/抖音读书口播稿,并输出项目可用的 raw JSON。只要用户提到“读书解说”“拆书稿”“视频号口播”“把这本书做成短视频”“长文档转视频脚本”“生成标题/封面/金句/raw JSON”,都应使用此 skill。

lzcjsyr/Book2Video

Cody的公众号长文写作skill。当用户需要撰写公众号文章、写稿子、续写文章、根据素材产出长文时使用。触发词包括但不限于:写文章、写稿子、帮我写、续写、扩写、公众号文章、长文、出稿、按我的风格写。即使用户只是说"帮我把这个写成文章"或"用我的风格写一下",只要上下文涉及内容创作和公众号输出,都应该触发。也适用于用户丢过来一个PDF、brief、新闻链接、语音转文字或任何素材说"帮我写篇文章"的场景。不要用于短内容(小红书帖子、推特、朋友圈)或纯标题摘要生成。

lzcjsyr/Book2Video

将中短篇专业报告、公司财报、券商研报、招股书、宏观经济分析及含有较多表格和数据的PDF文档,转化为高品质、数据精准的中文视频号/抖音口播解说稿。只要用户提到“研报”、“财报”、“商业分析报告”或需要“解读数据”、“把报告/PDF做成短视频”,务必使用此 Skill。

lzcjsyr/Book2Video

简化版的文稿生成技能。能够直接读取图书/文档,提取核心论点,生成高品质、口语化的视频号/抖音读书解说文稿并输出标准 JSON 格式。

lzcjsyr/Book2Video

The fallback workflow for authoring custom HyperFrames video compositions at any length or format — longer or multi-scene pieces, brand / sizzle reels, montages, title cards, static loops, and freeform compositions. Input- and length-agnostic. If a specialized workflow clearly fits the input — a marketed product, a website, a topic explainer, a GitHub PR, existing footage, a short motion graphic, or a Remotion port — prefer it (see /hyperframes); use this only as the general fallback when none fit.

相关技能