Knowledge Connector
Knowledge Connector should feel like a product line, not another graph utility.
Its job is not just to extract concepts. Its job is to help the user:
- - import notes and documents with low friction
- search across multiple documents from one query
- visualize concept relationships in a way that is easy to inspect
- get actionable graph results such as what to connect, review, or expand next
What This Skill Optimizes For
Default toward five high-value outcomes:
- - fast document import
- guided import onboarding
- cross-document knowledge retrieval
- relationship-aware graph views
- actionable next steps
Avoid drifting into “yet another adjacent knowledge skill”.
Primary Workflows
1. Import Experience
Use kc import-docs when the user wants to build a graph from multiple files or a notes directory.
Use kc import-wizard when the user wants a preview-first onboarding flow.
Good import behavior means:
- - accept files or a directory
- preserve source titles and paths
- show how many documents, concepts, and relations were created
- keep the user oriented after import
2. Cross-Document Search
Use kc search or kc query when the user asks:
- - where an idea appears across notes
- which documents mention a concept
- what concepts connect several documents
Results should show:
- - matching concepts
- matching source documents
- useful next actions
3. Relationship Visualization
Use kc visualize for full graph export and kc map for a concept-centered actionable subgraph.
Visualization should help the user answer:
- - what is central
- what is weakly connected
- what deserves review
4. Actionable Results
Do not stop at “here is the graph”.
The output should usually recommend one or more actions such as:
- - import more source material
- auto-connect newly imported concepts
- inspect a concept-centered subgraph
- verify weak relationships from source documents
- export a graph view for sharing or review
Core Commands
Import
CODEBLOCK0
Search
CODEBLOCK1
Map And Visualize
CODEBLOCK2
Manage
CODEBLOCK3
Output Standard
When the skill returns results, prefer this structure:
What Matched
Show concepts and source coverage.
Why It Matters
Explain the meaningful relationship or pattern.
Next Step
Tell the user what to do next with the graph.
Product Positioning
Knowledge Connector is strongest when the user has:
- - a growing notes corpus
- repeated concepts spread across files
- a need to move from storage to understanding
It is weaker if it only acts like a raw extractor with no import flow, no source-aware search, and no next-step guidance.
知识连接器
知识连接器应像一条产品线，而非又一个图谱工具。
它的职责不仅是提取概念，更是帮助用户：
- - 以低门槛方式导入笔记和文档
- 通过一次查询在多个文档中搜索
- 以易于检视的方式可视化概念关系
- 获得可操作的图谱结果，例如下一步应连接、审阅或扩展的内容
本技能优化目标
默认聚焦五大高价值成果：
- - 快速文档导入
- 引导式导入流程
- 跨文档知识检索
- 关系感知图谱视图
- 可操作的后续步骤
避免沦为又一个相近的知识技能。
主要工作流
1. 导入体验
当用户希望从多个文件或笔记目录构建图谱时，使用 kc import-docs。
当用户希望采用预览优先的引导流程时，使用 kc import-wizard。
良好的导入行为意味着：
- - 接受文件或目录
- 保留源标题和路径
- 显示创建的文档、概念和关系数量
- 导入后保持用户方向感
2. 跨文档搜索
当用户提出以下问题时，使用 kc search 或 kc query：
- - 某个想法出现在哪些笔记中
- 哪些文档提到了某个概念
- 哪些概念连接了多个文档
结果应显示：
3. 关系可视化
使用 kc visualize 进行完整图谱导出，使用 kc map 获取以概念为中心的可操作子图。
可视化应帮助用户回答：
4. 可操作结果
不要止步于这是图谱。
输出通常应推荐一个或多个操作，例如：
- - 导入更多源材料
- 自动连接新导入的概念
- 检视以概念为中心的子图
- 验证源文档中的弱关系
- 导出图谱视图用于分享或审阅
核心命令
导入
bash
kc import-wizard --dir notes/
kc import-docs --dir notes/
kc import-docs --files a.md b.md c.txt
搜索
bash
kc search 机器学习
kc answer 哪些文档把强化学习和规划连在一起？
kc query transformer --sources
kc query --ask 哪些文档同时提到了强化学习和规划？
映射与可视化
bash
kc map --concept 人工智能 --depth 2
kc visualize --format html --output graph.html
kc visualize --concept 机器学习 --depth 2 --output ml-graph.html
管理
bash
kc stats
kc export --output backup.json
kc import --file backup.json
输出标准
当技能返回结果时，优先采用以下结构：
匹配内容
展示概念和源文档覆盖率。
重要性说明
解释有意义的关系或模式。
下一步操作
告知用户接下来应如何操作图谱。
产品定位
知识连接器在以下场景中表现最佳：
- - 用户拥有不断增长的笔记库
- 重复概念分散在多个文件中
- 用户需要从存储转向理解
如果它仅充当原始提取器，缺乏导入流程、源感知搜索和下一步指导，其价值就会减弱。
