k2f-conference-handout
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseK2F Conference Handout — AI-Native Aesthetic Design
K2F 会议讲义 — AI 原生美学设计
Purpose
目的
Teach agents to author for attendee reference: what to keep from the talk, how to layer it for reading on paper or PDF, and how to express design intent (tone, hierarchy, density)—not how to build or export K2F packages.
content_and_design.md教会 Agent 编写 ,用于与会者参考:讲清应从演讲中保留什么、如何为纸质或 PDF 阅读分层组织内容,以及如何表达设计意图(基调、层级、信息密度)——而不是讲解如何构建或导出 K2F 包。
content_and_design.mdBuilt on K2F
基于 K2F 构建
This skill produces conference handouts as packages. Follow the k2f skill for authoring: unpack or init a workspace, edit the semantic tree, pack/verify, and export. K2F keeps agents on meaning (sections, takeaways, figures)—not pixel coordinates—while the engine compiles layout, typography, and theme into consistent output across PDF and other formats.
.K2FBefore you build, install the core skill: .
npx skills add suzheng/k2f --skill k2f本 skill 会将会议讲义生成为 包。按照 k2f skill 进行创作:解包或初始化工作区,编辑语义树,打包/验证并导出。K2F 让 Agent 专注于含义(章节、要点、图表),而不是像素坐标;引擎则将布局、排版和主题编译成在 PDF 及其他格式中保持一致的输出。
.K2F构建前,请先安装核心 skill:。
npx skills add suzheng/k2f --skill k2fWhen to Use
使用时机
- After a talk is outlined or accepted; before or after slides exist.
- When the user wants something attendees keep, not a CFP, not full slide export, not presenter notes.
- Print or PDF distribution at workshops, meetups, or conference sessions.
- 演讲已完成大纲规划或已获录用后;无论幻灯片是否已经制作完成,均可使用。
- 用户需要与会者会保存的内容,而不是 CFP、完整的幻灯片导出或演讲者备注。
- 用于工作坊、meetup 或 conference session 的打印与 PDF 分发。
Workflow
工作流
Step 1. Generate the content_and_design.md
第 1 步:生成 content_and_design.md
One blueprint pairing semantic content with aesthetic intent. Use labeled blocks the k2f skill can map to pages later.
Sub-steps
-
Session lock — Event, session title, slot length, audience job and situation (not “developers”), vendor-neutral vs vendor-run tone, and whether product names are allowed. State what the handout must do: follow-up reference, not marketing.
-
Source spine — List the talk in running order with minute estimates (two levels). Mark each beat Keep, Compress, or Drop. Handouts mirror the arc, not the slide count.
-
Attendee layers — Structure content in four tiers:
- Glance (10 s): title, speaker, event/date, optional QR or URL.
- Scan (1 min): 2–4 verb-first takeaways scoped to slot length (lightning: one idea, one takeaway; standard: one problem, two or three actions).
- Read (5–10 min): condensed sections aligned to the spine—hooks, diagrams, tables, code the room could not absorb live.
- Later: links, repos, reading, prerequisites, contact—short, sourced from the user.
-
Progressive beats — Where the live talk used reveals or demos, specify the static equivalent in the design doc (numbered steps, before/after pair, screenshot description, final code state). Never leave a section that only made sense animated.
-
Cut pass — Remove pitch-first openings, book-sized scope, speaker-journey framing, and “understand/appreciate” bullets. Replace with attendee-side sentences and actions they can perform Monday.
-
Design intent — Document tone, hierarchy, density, and palette for the engine (no JSON):
- Tone: attendee-facing; problem → approach → payoff; evidence and numbers the speaker can source.
- Hierarchy: takeaways and section titles carry the story; body supports; footnotes and links recede.
- Density: tighter than slides, looser than a paper—bullets and callouts over long paragraphs; cap ~800–1,200 words unless the user sets a limit.
- Palette: calm professional base, one accent for takeaways and warnings; high contrast for print; code and figures called out as full-width or inline.
Rules & constraints
- One angle per handout: one problem and one through-line; split multi-topic talks into separate handouts.
- Takeaway test: every takeaway starts with an action verb and fits the slot; eight takeaways in twenty-five minutes fails the test.
- Attendee voice: rewrite “I’ll show our journey” into what the reader gains.
- Honesty gate: no invented benchmarks, customers, or citations; mark gaps the talk skipped.
- Static-only assets: video, iframe, or live demo → screenshot, steps, or repo link in the blueprint.
Common failure modes
| Failure | Fix |
|---|---|
| Slide dump | Keep spine labels; drop title-only slides; expand only high-value beats. |
| Presenter notes pasted | Separate optional “facilitator” material; default doc is attendee-only. |
| Missing reveal content | Add final-state diagram, code, or numbered list for each progressive beat. |
| Pitch in disguise | Lead with the problem; product is the setting, not the headline. |
| Interchangeable title | Add system, scale, or tension from the talk. |
| Wall of code | One exemplar snippet plus link; annotate lines that matter. |
| Generic takeaways | Verb + object + context (“Audit egress before enabling cross-region replication”). |
Quality checklist
- Session lock and audience are explicit; tone matches event type.
- Spine mapped with Keep/Compress/Drop; one clear angle.
- 2–4 verb-first takeaways match slot length.
- Every live-only beat has a static substitute described.
- No pitch-first or speaker-centric openings; at least one sourced number or concrete system.
- Word budget stated; links and prerequisites in the Later layer.
- Design intent covers tone, hierarchy, density, and palette.
- All claims traceable to user-supplied material.
创建一份将语义内容与美学意图结合的蓝图。使用带标签的区块,便于 k2f skill 后续映射到页面。
子步骤
-
Session 定稿 — 记录活动、演讲标题、session 时长、受众的岗位和具体处境(不要只写“开发者”)、厂商中立或厂商主导的语气,以及是否允许出现产品名称。明确讲义必须发挥的作用:作为后续参考,而非营销材料。
-
内容主线(Source spine) — 按实际讲述顺序列出演讲内容,并提供两级时长估算。将每个内容节点标记为保留(Keep)、压缩(Compress)或删除(Drop)。讲义应映射演讲脉络,而不是幻灯片数量。
-
与会者阅读层级 — 将内容划分为四层:
- 速览(10 秒):标题、演讲者、活动/日期,以及可选的 QR 或 URL。
- 扫读(1 分钟):根据 session 时长安排 2–4 个以动词开头的要点(闪电演讲:一个想法、一个要点;标准演讲:一个问题、两到三个行动)。
- 阅读(5–10 分钟):与主线对应的精简章节——包括现场来不及消化的引入点、图表、表格和代码。
- 后续:链接、repo、阅读材料、先决条件和联系方式——内容简短,且来源于用户提供的信息。
-
渐进式节点 — 对于现场演讲中通过逐步揭示(reveal)或 demo 呈现的内容,在设计文档中说明相应的静态替代表达(编号步骤、前后对比、截图描述、最终代码状态)。绝不能留下只有动态演示才能理解的部分。
-
删减处理(Cut pass) — 删除以推销开场的段落、书籍规模的范围、演讲者个人历程式框架,以及“了解/欣赏”式要点。改写为站在与会者角度的陈述,以及他们周一就能执行的行动。
-
设计意图 — 为引擎记录基调、层级、信息密度和配色(不使用 JSON):
- 基调(Tone):面向与会者;按问题 → 方法 → 收益展开;使用演讲者能够提供来源的证据和数字。
- 层级(Hierarchy):要点和章节标题承载叙事;正文提供支撑;脚注与链接弱化呈现。
- 密度(Density):比幻灯片更紧凑,但比论文更疏朗;用项目符号和 callout 代替长段落;除非用户设定上限,否则控制在约 800–1,200 字。
- 配色(Palette):以平静专业的基础色为主,用一种强调色突出要点和警示信息;针对打印保证高对比度;代码和图表以通栏或嵌入式形式突出显示。
规则与约束
- 一份讲义一个角度:只聚焦一个问题和一条主线;将包含多个主题的演讲拆分为独立讲义。
- 要点测试:每个要点都以行动动词开头,并与 session 时长相符;25 分钟演讲包含 8 个要点即无法通过测试。
- 与会者口吻:将“我会展示我们的历程”改写为读者能够获得什么。
- 真实性门槛:不得虚构 benchmark、客户或引用;明确标记演讲中跳过的信息缺口。
- 仅使用静态资产:视频(video)、iframe 或现场 demo → 在蓝图中改为截图、步骤或 repo 链接。
常见失败模式
| 失败模式 | 修复方式 |
|---|---|
| 幻灯片内容倾倒 | 保留主线标签;删除只有标题的幻灯片;仅扩展高价值节点。 |
| 直接粘贴演讲者备注 | 将可选的“引导者”材料单独放置;默认文档仅供与会者使用。 |
| 缺少揭示后的最终内容 | 为每个渐进式节点添加最终状态图、代码或编号清单。 |
| 变相推销 | 从问题切入;产品只是背景,不是标题。 |
| 标题缺乏辨识度 | 从演讲中加入系统、规模或张力。 |
| 大段代码墙 | 保留一段代表性代码并附链接;标注关键代码行。 |
| 泛泛的要点 | 动词 + 对象 + 情境(例如“在启用跨区域复制前检查出口流量”)。 |
质量检查清单
- Session 定稿和受众清晰明确;语气与活动类型一致。
- 主线已按保留/压缩/删除映射;只有一个清晰角度。
- 2–4 个以动词开头的要点与 session 时长匹配。
- 每个仅在现场呈现的节点都描述了静态替代内容。
- 没有以推销或演讲者为中心的开场;至少包含一个有来源的数字或具体系统。
- 明确字数预算;链接和先决条件置于“后续”层。
- 设计意图涵盖基调、层级、密度和配色。
- 所有主张均可追溯至用户提供的材料。
Step 2. Use the content_and_design.md and k2f skill to generate a K2F file and export to an appropriate file format.
第 2 步:使用 content_and_design.md 和 k2f skill 生成 K2F 文件并导出为适当的文件格式
Follow k2f to instantiate pages from the blueprint, apply roles from design intent, verify, and export the attendee-ready handout (typically PDF for print or sharing).
按照 k2f 的流程,从蓝图实例化页面,应用设计意图中的各项角色,验证并导出可直接分发的讲义(通常为便于打印或分享的 PDF)。