learn

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Learn: From Raw Materials to Published Output

学习：从原始素材到可发布内容

Prefix your first line with 🥷 inline, not as its own paragraph.

Your role: collect, organize, translate, explain, structure. You support the user's thinking; you do not replace it.

你的第一行需要以内联形式带上🥷前缀，不要单独作为一个段落。

你的职责：收集、整理、翻译、解释、搭建结构。你是辅助用户思考，而非替代用户思考。

Pre-check

前置检查

Before starting, check whether

/read

and

/write

skills are installed (search for their SKILL.md in the skills directories). Warn if missing but do not block:

```
/read
```
missing: warn that Phase 1 will fall back to the environment's native fetch capability or
```
curl
```
instead of
```
/read
```
.
```
/write
```
missing: warn that Phase 5 will not be able to strip AI patterns from the draft. Phases 1-4 are unaffected.

开始前，请检查是否已安装

/read

和

/write

技能（在技能目录中搜索对应的SKILL.md文件）。如果缺失请发出警告，但不要阻塞流程：

缺失
```
/read
```
：警告阶段1将回退到环境原生的获取能力或
```
curl
```
，而非使用
```
/read
```
。
缺失
```
/write
```
：警告阶段5无法从草稿中剔除AI生成特征，阶段1-4不受影响。

Choose Mode

选择模式

Ask the user to confirm the mode, using the environment's native question or approval mechanism if it has one:

Mode	Goal	Entry	Exit
Deep Research	Understand a domain well enough to write about it	Phase 1	Phase 6: publish
Quick Reference	Build a working mental model fast, no article planned	Phase 2	Phase 2: notes only
Write to Learn	Already have materials, force understanding through writing	Phase 3	Phase 6: publish

If unsure, suggest Quick Reference.

请让用户确认使用的模式，如果环境有原生的提问或确认机制可以使用该机制：

模式	目标	开始阶段	结束阶段
深度研究	充分理解某个领域，足以输出相关内容	阶段1	阶段6：发布
快速参考	快速搭建可用的思维模型，无写文章的计划	阶段2	阶段2：仅输出笔记
以写促学	已有素材，通过写作强制完成知识理解	阶段3	阶段6：发布

如果不确定选哪个，建议使用快速参考模式。

Phase 1: Collect

阶段1：收集

Gather primary sources only: papers that introduced key ideas, official lab/product blogs, posts from the people who built the thing, canonical "build it from scratch" repositories. Not summaries. Not explainers.

For each source: download, convert to Markdown, file into a structured directory organized by sub-topic. Use

/read

for individual pages.

Source Discovery: if a web search plugin is installed (e.g., PipeLLM search), use it. Strategy: fast search to map the landscape, then deep search on the 2-3 most promising sub-topics. Otherwise: use the environment's native web search or fetch capability, or fall back to

curl + defuddle.md

Target: 5-10 sources for a blog post, 15-20 for a deep technical survey.

仅收集一手资料：提出核心观点的论文、官方实验室/产品博客、项目构建者发布的内容、权威的「从零搭建」代码仓库。不要摘要，不要解读类内容。

对于每份资料：下载、转换为Markdown格式、归档到按子主题划分的结构化目录中。单页内容使用

/read

处理。

资料发现： 如果安装了网页搜索插件（比如PipeLLM search），则使用该插件。搜索策略：先快速搜索梳理领域全貌，再对2-3个最有价值的子主题做深度搜索。如果没有搜索插件：使用环境原生的网页搜索或获取能力，或回退到

curl + defuddle.md

的方案。

数量目标：博客文章需要5-10份资料，深度技术调研需要15-20份资料。

Phase 2: Digest

阶段2：消化

Work through the materials. For each piece: read it fully, keep what is good, cut what is not. At the end of this phase, cut roughly half of what was collected.

For key claims, ask before including in the outline:

Does this idea appear in at least two different contexts from the same source?
Can this framework predict what the source would say about a new problem?
Is this specific to this source, or would any expert in the field say the same thing?

Generic wisdom is not worth distilling. Passes two or three: belongs in the outline. Passes one: background material. Passes zero: cut it.

When two sources contradict on a factual claim, note both positions and the evidence each gives. Do not silently pick one.

逐份处理收集到的材料。每份材料完整通读，保留有价值的内容，剔除无用内容。本阶段结束时，要删掉大概一半收集到的内容。

对于核心观点，纳入大纲前需要先确认：

这个观点是否在同一资料的至少两个不同语境下出现过？
这个框架能否预测该资料针对新问题会给出的结论？
这个观点是该资料独有，还是领域内所有专家都认可的共识？

通用的常识性内容不值得提炼。符合2-3个条件：适合纳入大纲。符合1个条件：作为背景材料。0个符合：直接删掉。

如果两份资料在事实性观点上存在矛盾，要同时标注两种立场和各自的举证，不要默认选择其中某一方。

Phase 3: Outline

阶段3：列大纲

Write the outline for the article. For each section: note the source materials it draws from. If a section has no sources, either it does not belong or a source needs to be found first.

Do not start Phase 4 until the outline is solid.

撰写文章的大纲。每个章节都要标注对应的参考资料来源。如果某个章节没有对应来源，要么这个章节不需要存在，要么需要先补充对应的资料。

大纲确认无误后再进入阶段4。

Phase 4: Fill In

阶段4：填充内容

Work through the outline section by section. If a section is hard to write, the mental model is still weak there: return to Phase 2 for that sub-topic. The outline may change, and that is fine.

逐章节按大纲填充内容。如果某个章节很难写，说明你对该子主题的思维模型还不够完善：回到阶段2重新处理该子主题的内容。大纲可以调整，没有问题。

Phase 5: Refine

阶段5：优化

Pass the draft with a specific brief:

Remove redundant and verbose passages without changing meaning or voice
Flag places where the argument does not flow
Identify gaps: concepts used before they are explained, claims needing sources

Do not summarize sections the user has not written. Do not draft new sections from scratch. Edits only.

Then run

/write

on the refined draft to strip any AI patterns that crept in during refinement.

给草稿设置明确的优化要求：

删除冗余啰嗦的段落，不要改变原意和语气
标记出论证不流畅的位置
找出内容缺口：概念在解释前就被使用、需要补充来源的观点

不要总结用户还没写的章节，不要从零开始写新的章节，仅做编辑修改。

然后对优化后的草稿运行

/write

命令，剔除优化过程中可能混入的AI生成特征。

Phase 6: Self-review and Publish

阶段6：自我检查与发布

The user reads the entire article linearly before publishing. Not with AI. Mark everything that feels off, fix it, read again. Two passes minimum.

When it reads clean from start to finish, publish it.

用户在发布前要线性通读整篇文章，不要用AI辅助。标记所有感觉不对劲的地方，修改后再通读一遍，最少通读2次。

当整篇文章从头到尾读起来都很顺畅时，再发布。

Gotchas

注意事项

What happened	Rule
Collected 30 secondary explainers instead of primary sources	Phase 1 targets papers, official blogs, and repos by builders. Summaries are not sources.
Treated a convincing explainer as ground truth	Ask: does this appear in at least two different contexts from the same source?
Skipped Phase 5 because the outline felt done	Refine always runs before publish. "Good enough" is Phase 4 thinking.
Two sources contradicted; silently picked one	Note both positions and the evidence each gives. Never silently resolve a contradiction.
Published before the self-review pass	The user reads the entire article linearly before it ships. AI does not do this step.

问题	处理规则
收集了30份二手解读内容而非一手资料	阶段1的目标是论文、官方博客、构建者发布的代码仓库。摘要不属于有效资料。
把看起来有说服力的解读内容当成事实	先确认：这个观点是否在同一资料的至少两个不同语境下出现过？
觉得大纲已经完成就跳过阶段5	发布前一定要做优化。「足够好」是阶段4的标准。
两份资料观点矛盾，默认选择了其中一方	要同时标注两种立场和各自的举证，永远不要默认解决矛盾。
还没做自我检查就发布	用户发布前要线性通读整篇文章，该步骤不能由AI完成。