capture-learning

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Capture Learning

记录学习成果

This skill captures and documents technical realizations and learnings directly into the user's Logseq vault (

/Users/hw/logseq

此 skill 会将技术认知和经验直接记录到用户的 Logseq vault（

/Users/hw/logseq

）。

Trigger Conditions

触发条件

Manual: The user explicitly says "document this", "capture this learning", "save this realization", "TIL", or similar.
Automatic — Proactively capture (then confirm) when the user:
- Shares a debugging root cause that turned out to be non-obvious (e.g., "turns out
```
ignore_changes
```
  doesn't apply to computed attributes")
- Expresses surprise or correction about how something works ("I didn't know", "oh interesting", "huh", "good to know", "wait so X actually…")
- Discovers a CLI flag, config option, or tool behavior they hadn't encountered before
- Has a mental model shift — their understanding of a system fundamentally changes
- Resolves a confusing bug and the fix reveals how the system actually behaves
In these cases, capture the learning first, then say something like: "I've saved this to your Logseq vault — let me know if you'd like to adjust the wording."

手动：用户明确说出 "document this"、"capture this learning"、"save this realization"、"TIL" 或类似表述。
自动——在以下情况下主动记录（然后确认）：
- 用户分享了一个最终发现并不显而易见的调试根因（例如，“原来是
```
ignore_changes
```
  不适用于计算属性”）
- 用户对某项运作方式表达惊讶或纠正（“我不知道”、“哦，有意思”、“原来如此”、“原来如此”、“等等，所以 X 实际上……”）
- 发现此前未曾接触的 CLI flag、配置选项或工具行为
- 心智模型发生转变——对某个系统的理解发生根本变化
- 解决了一个令人困惑的 bug，而修复过程揭示了系统的实际行为
在这些情况下，先记录学习内容，然后说：“我已将其保存到你的 Logseq vault——如果你想调整措辞，请告诉我。”

When NOT to trigger

不应触发的情况

Routine changes with no new insight (simple refactors, typo fixes, straightforward tasks)
Information the user clearly already knew and was just explaining to you
Secondhand info — something a colleague or someone else discovered, not the user
Generic best practices that aren't specific to the user's stack or project
The user is asking a question (not sharing a realization)

没有新见解的常规变更（简单重构、错别字修复、直接明了的任务）
用户显然早已知道、只是在向你解释的信息
二手信息——由同事或其他人发现，而非用户本人
与用户技术栈或项目无关的通用最佳实践
用户在提问（而非分享认知）

Execution Workflow

执行工作流

When capturing a learning:

Summarize the Learning: Formulate a concise, clear explanation of the realization (what it is, why it works that way, and any context or code snippets that illustrate it). The top line of the entry must focus on meaningful text describing the learning, rather than starting with tags. Place any relevant tags at the end of the line or in nested blocks.
Search for Related Content (Namespaces Preferred): Before creating a new page or defaulting to the daily journal, use search tools (
```
grep_search
```
,
```
glob
```
) to check the
```
/Users/hw/logseq
```
vault for existing, relevant pages. Look for matching namespaces (e.g., existing
```
learnings___
```
or
```
projects___
```
pages related to the topic).
Apply Logseq Conventions: You MUST activate and use the
```
logseq-organization
```
skill to format and place the note correctly based on the search results.
- If related pages exist, append the learning to the most relevant existing page under an appropriate heading.
- If no relevant page exists and the learning is small/isolated, add it to today's daily journal under the
```
- ## #learnings
```
  heading.
- If no relevant page exists and the learning is substantial or forms a comprehensive new topic, ask the user if they'd like to create a specific learning page (e.g.,
```
pages/learnings___<topic-name>.md
```
  ).
Link Source Context: ALWAYS link the relevant project page (e.g.,
```
[[projects/personal/saalnaama]]
```
) when capturing a learning, especially in the daily journal. This provides essential context for where the learning originated.
Write to Vault: Use file writing tools to append or write the formatted entry to the chosen file in
```
/Users/hw/logseq
```
.
Confirm: Provide the user with a brief confirmation and a snippet of what was saved.

记录学习内容时：

总结学习内容：用简洁、清晰的方式表述该认知（它是什么、为何会这样运作，以及任何有助于说明的上下文或代码片段）。条目的首行必须以描述该学习成果的实质性文本为主，不能以标签开头。将相关标签放在该行末尾或嵌套块中。
搜索相关内容（优先使用 Namespaces）：在创建新页面或默认使用 daily journal 之前，使用搜索工具（
```
grep_search
```
、
```
glob
```
）检查
```
/Users/hw/logseq
```
vault 中是否已有相关页面。查找匹配的 namespaces（例如，与该主题相关的现有
```
learnings___
```
或
```
projects___
```
页面）。
遵循 Logseq 规范：必须激活并使用
```
logseq-organization
```
skill，根据搜索结果正确设置条目的格式和位置。
- 如果相关页面存在，将学习内容追加到最相关的现有页面中合适的标题下。
- 如果不存在相关页面且学习内容较小或独立，将其添加到今日 daily journal 的
```
- ## #learnings
```
  标题下。
- 如果不存在相关页面，且学习内容内容详尽或形成一个全面的新主题，则询问用户是否愿意创建特定学习页面（例如
```
pages/learnings___<topic-name>.md
```
  ）。
链接来源上下文：记录学习内容时，始终链接相关项目页面（例如
```
[[projects/personal/saalnaama]]
```
），尤其是在 daily journal 中。这可以提供学习来源的关键上下文。
写入 Vault：使用文件写入工具，将格式化后的条目追加到或写入
```
/Users/hw/logseq
```
中选定的文件。
确认：向用户提供简短确认和已保存内容的片段。

Examples of Learnings

学习内容示例

"Terraform write-only resources for secrets don't store state in the clear."
"Drupal canvas stores its content as base64-encoded strings in a specific database table."
"Kubernetes VerticalPodAutoscaler evicts pods to apply new resource requests."

“Terraform write-only resources 不会以明文形式存储 state。”
“Drupal canvas 会将其内容以 base64 编码的字符串形式存储在数据库中的特定表中。”
“Kubernetes VerticalPodAutoscaler 会驱逐 pods，以应用新的资源请求。”