ie-connect-dots

Reasoning Layer: Idea Connection and Pattern Discovery

This skill is responsible for analyzing the relationships among multiple notes or ideas, discovering implicit connections and long-term patterns. This is the middle layer in the three-layer architecture of the Inspiration Engine, receiving retrieval results from the memory layer and providing structured reasoning outputs for the insight layer.

Applicable Scenarios

Use when it is necessary to understand the relationships among multiple notes or ideas. Typical scenarios include:

• The user provides multiple notes and wants to know what connections exist among them
• Invoked by the ie-engine orchestrator to process the retrieval results of ie-retrieve-memory
• The user wants to discover the thematic patterns they have been focusing on over the long term
• The user wants to classify and organize a group of scattered ideas

Inapplicable Scenarios

The following scenarios should not use this skill (to avoid false triggers):

• Retrieving or searching notes (use ie-retrieve-memory)
• Generating insight text or exploration suggestions (use ie-generate-insight)
• Content analysis or summarization of a single note (can be done directly)
• General data analysis or statistics (use other appropriate tools)

---

Three Reasoning Capabilities

Capability One: Semantic Clustering

Group multiple notes or ideas by semantic similarity.

Analysis Method:

1. Extract the core theme and key concepts of each note
2. Divide the notes into several groups based on thematic similarity
3. Name each group, summarizing the common theme of the notes within the group
4. Annotate the near-far relationships between groups

Output Format:

语义聚类结果
---

聚类 A：[聚类名称]
  核心主题：[一句话描述]
  包含笔记：
    - [笔记标题1]：[与聚类主题的关联点]
    - [笔记标题2]：[与聚类主题的关联点]

聚类 B：[聚类名称]
  核心主题：[一句话描述]
  包含笔记：
    - [笔记标题3]：[与聚类主题的关联点]

聚类间关系：
  聚类A 与 聚类B：[关系描述，例如"A是B的技术基础"]

Capability Two: Connection Hypotheses

Propose possible connections between different ideas. This is the most core reasoning capability of the Inspiration Engine.

Analysis Method:

1. Identify ideas from different clusters or different time periods
2. Look for potential associations between them (shared concepts, complementarity, causality, evolutionary relationships)
3. Provide a confidence level and reasoning basis for each connection hypothesis
4. Prioritize discovering those connections that are "not obvious but have potential"

Connection Type Reference:

• Complementary type: two ideas can be combined together to form a more complete solution
• Evolutionary type: one idea is a natural extension or upgrade of another idea
• Causal type: the conclusion of one idea may be the premise of another idea
• Analogical type: ideas from two different domains have similar structures or patterns
• Conflict type: two ideas contradict each other and need to be reconciled or chosen between

Output Format:

连接假设
---

假设 1：[简短标题]
  连接类型：[互补型/演化型/因果型/类比型/冲突型]
  想法 A：[笔记标题或想法描述]
  想法 B：[笔记标题或想法描述]
  连接推理：[为什么认为这两个想法有关联，2-3句话]
  置信度：高/中/低
  潜在价值：[如果这个连接成立，可能带来什么]

假设 2：[简短标题]
  ...

Capability Three: Long-Term Pattern Recognition

Identify recurring themes and focus trends from notes spanning a relatively long time period.

Analysis Method:

1. Arrange the notes along a timeline
2. Annotate the core theme of each note
3. Identify themes that recur across multiple time periods
4. Analyze the evolution trends of themes (strengthening, weakening, shifting)

Output format:

长期模式
---

模式 1：[主题名称]
  出现频率：在 [时间范围] 中出现 [次数] 次
  演变趋势：[增强/减弱/稳定/转向]
  代表笔记：
    - [时间] [笔记标题]
    - [时间] [笔记标题]
  趋势解读：[一句话解读这个模式意味着什么]

模式 2：[主题名称]
  ...

---

Input requirements

This skill accepts input in any of the following forms:

1. Retrieval results from ie-retrieve-memory(recommended, the standard pipeline path)
   • Includes note titles, IDs, tags, and content summaries
2. A list of note contents or ideas provided directly by the user
   • At least 2 notes or ideas are required to perform relationship analysis
3. A list of note IDs
   • The note contents need to be read first before analysis

If the input contains fewer than 2 items, prompt the user to provide more material or to first use ie-retrieve-memory for retrieval.

---

Comprehensive output structure

When all three capabilities are used together, organize the complete output according to the following structure. This format is designed to be passed directly to the insight layer (ie-generate-insight) for use.

推理分析结果
===

一、语义聚类
[聚类结果]

二、连接假设
[连接假设列表，按置信度排序]

三、长期模式
[模式识别结果]

---
分析置信度：高/中/低
输入笔记数量：[数量]
分析覆盖时间范围：[起止时间]

---

Relationship with other inspiration engine skills

This skill is the inspiration engine's reasoning layer, occupying the middle position in the three-layer architecture:

ie-generate-insight（洞见层）<-- 消费推理结果
ie-connect-dots（推理层）   <-- 本技能，分析想法关系
ie-retrieve-memory（记忆层）<-- 提供原始素材

• Upstream: ie-retrieve-memory (memory layer) provides the retrieved notes
• Downstream: ie-generate-insight (insight layer) receives this skill's reasoning results
• Orchestration: ie-engine is responsible for chaining the calls

This skill can be used independently (the user directly provides notes for you to analyze relationships) or as part of the inspiration engine pipeline.