• ❌ No admission audit (does not judge whether it qualifies for storage)
• ❌ No content processing (does not extract cognitive frameworks)
• ✅ Focus on golden sentence extraction
• ✅ Suitable for quickly accumulating writing materials

1. The user explicitly requests "extract golden sentences", "excerpt good sentences", "save these expressions"
2. The user says "This sentence is well-written", "This metaphor is brilliant" while reading an article
3. The user enters the shortcut command:

   /extract-quotes

   or

   /extract-quotes [file path]

• If the user needs to organize external materials and store them in the library → Use

  /kn

  (knowledge_auditor)
• If the user wants to learn the article structure → Use

  /analyze-writing

  (writing_analyzer)

1. Obtain Input:
   • If the user provides a file path: Read directly
   • If the user pastes content: Use the pasted text
   • If the user provides neither: Ask the user to provide a file path or paste content
2. Confirm Extraction Criteria:
   • Ask the user: "What type of golden sentences do you want to extract?"
     • Comparative Golden Sentences (A vs B type expressions)
     • Metaphor & Analogy (vivid metaphors for abstract concepts)
     • Counterintuitive Viewpoints (expressions that subvert conventional cognition)
     • Call to Action (inspiring concluding sentences)
     • All Types (default)

• ✅ Not a common expression (e.g., "We must work hard")
• ✅ Has a unique perspective or metaphor
• ✅ Stands out and impresses

• ✅ Can stand alone and be understood without the original context
• ✅ Conveys a complete meaning without context
• ✅ Suitable for direct use in one's own articles

• ✅ Has impact or resonance
• ✅ Makes readers pause and think
• ✅ Has a vivid imagery or rhythmic flow

1. Sentence-by-Sentence Scan:
   • Read the full text, identify sentences that meet the criteria
   • Retain the original expression (no rewriting)
   • Mark the paragraph position (for easy context retrieval)
2. Classify and Organize:
   markdown

   ## Comparative Golden Sentences
   - "Exams require existing answers and correctness; the market requires usefulness and problem-solving"
     [Source: Paragraph 3]
   
   ## Metaphor & Analogy
   - "Treat your current girlfriend the way you treated your ex, and you can imagine the result"
     [Source: Opening]
   
   ## Call to Action
   - "Don't learn first then act, act first then learn"
     [Source: Paragraph 5]

3. Quality Screening:
   • Keep only 3-10 sentences (avoid "hoarding")
   • Prioritize the most impactful ones
   • If more than 10 sentences are identified, ask the user "Do you want to save all of them?"

[Golden Sentence 1]

• Applicable Scenario: [When to use]
• Source Paragraph: [Original Paragraph X]

[Golden Sentence 2]

• Applicable Scenario: [When to use]
• Source Paragraph: [Original Paragraph X]

[Golden Sentence 3]

• Applicable Scenario: [When to use]
• Source Paragraph: [Original Paragraph X]

[Golden Sentence 4]

• Applicable Scenario: [When to use]
• Source Paragraph: [Original Paragraph X]

• Original Text: [Article Title]
• Extraction Date: [Today's Date]

---

1. Duplicate Check: Retrieve in the directory

   E:\OBData\ObsidianDatas\3通用技能\内容创作\Attachment(素材库)

   :
   • Keywords: topic keywords, similar expressions
   • Check if there are existing golden sentence notes on the same topic
2. Proposals:
   • Scenario A (Append Mode): If a golden sentence note on the same topic is found, suggest appending to the existing note (avoid fragmentation)
   • Scenario B (New Note Mode): If no inventory is found or the topic is completely different, create a new note
3. Archiving Location:

   E:\OBData\ObsidianDatas\3通用技能\内容创作\Attachment(素材库)\[Topic] Golden Sentence Materials.md

1. Ask for User Confirmation:
   • Display the extracted golden sentence list
   • Ask: "Do you want to keep all these golden sentences? Do you need to delete or add any?"
2. Execute Writing:
   • After user confirmation, physically write to the file
   • Return the file path
3. Follow-up Suggestions:
   • Remind the user to delete the source file (if extracted from an external article)
   • Suggest reviewing the golden sentence library regularly (once a month)

1. Restraint Principle: Strictly control the quantity (3-10 sentences) to avoid "material hoarding"
2. Original Flavor: Retain the original expression, no rewriting (unless requested by the user)
3. Scenario-based: Each golden sentence must be marked with "Applicable Scenario"
4. No Empty Talk: Must use the

   Write

   tool to physically write to the file

• Uniqueness: Is it not a common expression?
• Quotability: Can it be understood without the original context?
• Emotional Impact: Does it have impact or resonance?
• Applicability: Can it be used in writing next week?

• knowledge_auditor (

  /kn

  ): If the user needs a complete audit + storage process, transfer to it
• writing_analyzer (

  /analyze-writing

  ): If the user says "I want to learn the writing structure of this article" after extracting golden sentences, recommend using the writing analyzer
• thought-mining (

  /thought-mining

  ): If the user wants to write a new article based on the golden sentences, recommend using the thought mining assistant

1. Copyright Awareness: Remind the user to indicate the source when quoting
2. Avoid Hoarding: If the user frequently extracts but never uses, issue a warning: "You have accumulated XX golden sentences, but haven't used any in the last 3 months. Use them up before extracting more."
3. Regular Cleanup: Remind the user every quarter to clean up golden sentences that "have never been quoted"