C31 Curator

Information curation and document generation skill. Takes the output of C31-research (facts) and C31-storm (judgments) and organizes them into deliverable documents.

Not just "formatting". Curation includes:

• Selecting what to include based on audience and purpose
• Structuring information for maximum impact
• Adding narrative flow (not just bullet points)
• Ensuring all claims have sources
• Generating multiple formats (email, report, outline) from same input

When to Use

• "把调研整理成文档"
• "帮我写一封套磁邮件"
• "把这个分析写成投资备忘录"
• "生成论文大纲"
• "整理成可以发给别人的格式"
• After C31-storm: "能不能写成..."

When NOT to Use

• User just wants quick judgment (C31-storm is enough)
• No prior research context (run C31-research first)
• User explicitly says "不用整理，告诉我重点就行"

Core Principles

1. Audience-first — same content, different packaging for different readers
2. Source-preservation — every claim in final doc must trace back to research
3. Narrative flow — not a dump, a story with beginning, middle, end
4. Moderator function — check if any important information was unused in prior stages
5. Multiple formats — one input, many outputs (email, memo, outline, slide deck)

Pipeline

Input: C31-research findings + C31-storm synthesis + user intent (what document? for whom?)
  ↓
Phase 1: Purpose Clarification → what format? audience? tone?
  ↓
Phase 2: Content Selection → what to include, what to exclude, what to emphasize
  ↓
Phase 3: Moderator Check → what important info was unused in prior stages?
  ↓
Phase 4: Structure & Flow → organize into narrative with transitions
  ↓
Phase 5: Format Generation → produce the actual document
  ↓
Phase 6: Source Attachment → add citation list / traceability
  ↓
Output: Deliverable document + metadata

Phase 1: Purpose Clarification

Questions to resolve (ask user if unclear):

1. Format: email / memo / report / outline / slide deck / social media post?
2. Audience: who will read this? what do they care about? what's their prior knowledge?
3. Tone: formal / casual / persuasive / informative / provocative?
4. Length: one page / five pages / tweet-length?
5. Action: what should reader do after reading? (reply, approve, think, share?)

Default assumptions (if user doesn't specify):

• Format: markdown document
• Audience: user themselves (personal notes)
• Tone: matching user's style (check USER.md)
• Length: 1 page for summaries, 3-5 pages for reports
• Action: inform and recommend

Phase 2: Content Selection

From C31-research + C31-storm, select:

Exclusion rules:

• Remove technical details audience doesn't need
• Remove duplicate information
• Remove low-confidence claims unless caveated
• Remove sources that don't add to the narrative

Phase 3: Moderator Check

This is the Co-STORM Moderator function adapted for C31-curator.

After selecting content, ask:

1. What information from C31-research was NOT used in C31-storm?
   • Was it irrelevant, or was it overlooked?
   • If overlooked and important: flag for inclusion
2. What claims in C31-storm lack sufficient source support?
   • Flag for caveats or removal
3. What would change the conclusion if added?
   • If something was omitted that would significantly change the synthesis, note it
4. What's the strongest counter-argument not represented?
   • If all perspectives were "pro", is there a "con" missing?

Moderator output:

{
  "moderator_check": {
    "unused_but_relevant": ["info from research not used in analysis"],
    "unsupported_claims": ["claims needing better sources"],
    "omitted_game_changers": ["info that would change conclusion"],
    "missing_counterarguments": ["perspectives not represented"]
  }
}

If moderator finds issues:

• Minor: add caveats in final document
• Major: flag to re-run C31-storm with additional perspective
• Critical: recommend re-running C31-research

Phase 4: Structure & Flow

Document structures by type:

Email / Letter

Opening: hook (why this matters to recipient)
Body: 2-3 key points, each with evidence
Bridge: connect to recipient's interests
Ask: what do you want them to do?
Closing: professional sign-off

Memo / Report

Executive Summary: one-paragraph synthesis
Context: why this topic matters
Findings: facts with sources
Analysis: conflicts and judgments
Recommendations: actionable, prioritized
Appendix: full sources

Outline

Title
I. Main Point
   A. Supporting evidence
   B. Counter-point and response
II. Main Point
   ...

Slide Deck (text description)

Slide 1: Title + one-liner
Slide 2: Problem / Context
Slide 3: Key finding 1
Slide 4: Key finding 2
Slide 5: Conflict / Tension
Slide 6: Recommendation
Slide 7: Next steps / Ask

Flow principles:

• Start with what matters most to the audience
• Each paragraph transitions to the next (not a list of facts)
• Place the "ask" or "recommendation" where it has maximum impact
• End with momentum toward action

Phase 5: Format Generation

Generate the actual document in requested format.

For email:

• Subject line: specific, not generic (e.g., not "Research on Professor" but "Kyoto Application: Effectuation + Technical Leverage Angle")
• Body: 3-5 paragraphs max
• Signature: user's name + relevant credential

For report:

• Use markdown headers
• Bold key terms sparingly
• Tables for comparisons only
• Bullet points for lists, not paragraphs

For outline:

• Roman numerals for main sections
• Capital letters for sub-points
• Include word count estimates per section

For social media:

• Hook in first line
• Thread format if >280 chars
• One insight per post
• End with question or call to action

Phase 6: Source Attachment

All documents must include:

─── Sources ───
• [Claim 1] → [Source URL or reference]
• [Claim 2] → [Source URL or reference]
...

─── Confidence Notes ───
• High confidence: [list]
• Medium confidence: [list]
• Low confidence / speculative: [list]

Output Contract

{
  "phase": "curation",
  "document_type": "email|memo|report|outline|slides|social",
  "audience": "description",
  "tone": "formal|casual|persuasive|informative",
  "content_selection": {
    "included": [...],
    "excluded": [...],
    "emphasized": [...]
  },
  "moderator_check": {
    "unused_but_relevant": [...],
    "unsupported_claims": [...],
    "omitted_game_changers": [...],
    "missing_counterarguments": [...]
  },
  "document": "full text of the generated document",
  "sources": [...],
  "confidence": "high|medium|low"
}

Integration with C31-research → C31-storm → C31-curator

Full chain:

User asks: "调研一下X"
  → C31-research (facts)
  → conflict detected → C31-storm (judgments)
  → user says "写成文档" → C31-curator (deliverable)

Trigger words for C31-curator:

• "写成..."
• "整理成..."
• "生成..."
• "发给他..."
• "邮件..."
• "报告..."
• "大纲..."

Handoff from C31-storm:

• C31-storm passes: full synthesis + conflict map + source audit
• C31-curator asks: "what format?" or uses default
• C31-curator does NOT re-research; it reorganizes existing material

Moderator vs C31-curator: What's the Difference?

In C31-curator, the Moderator function is Phase 3 — a quality gate that checks if prior stages missed important information.

Example: Yamada Professor → Curation

Input: C31-storm synthesis + user says "帮我写一封套磁邮件"

Phase 1:

• Format: email
• Audience: Professor Yamada (academic, Japanese, values critical thinking)
• Tone: respectful but intellectually engaging, not obsequious
• Length: 3 paragraphs + signature
• Action: invite response / express interest in his research

Phase 2:

• Include: his latest critical turn, technical leverage gap, your background
• Exclude: detailed conflict map, source audit, blind spots about his career
• Emphasize: the frontier question about AI and Effectuation

Phase 3 (Moderator Check):

• Unused but relevant: his 2025 JOM paper on independent directors (shows methodological rigor)
• Unsupported claims: none
• Omitted game-changer: none
• Missing counterargument: "what if he thinks technical leverage is irrelevant to management?"

Phase 4-5: Generated Email

Subject: Kyoto Application: Technical Leverage and Effectuation

Dear Professor Yamada,

I am writing to express my interest in your research, particularly your recent work on the governance of management knowledge (経営学'の/と'権力). Your critique of how academic knowledge becomes marketized resonates deeply with my own experience as a technically-oriented founder.

Your research on effectuation has been foundational in Japan, yet I notice a gap that perhaps only someone with your critical perspective can address: when AI enables one person to build what previously required a team, does the effectuation question shift from "whom do I know" to "what can I build"? This is not merely a technical question—it is a question about the redistribution of entrepreneurial power, which seems directly connected to your critique of knowledge governance.

I would welcome the opportunity to explore this intersection under your guidance.

[Name]
[Background: technical founder, interested in entrepreneurship as personal freedom]

Phase 6: Sources attached

• Yamada (2026) 経営学'の/と'権力
• Yamada & Yoshikawa (2025) JOM
• J-Global researcher profile

Anti-Patterns