knowledge manager workflow

Knowledge Manager Workflow

Complete 6-phase workflow guide for content processing

Workflow Overview

Phase 0: Load Configuration
    ↓
Phase 1: Detect Input Source
    ↓
Phase 1.5: Collect User Preferences
    ↓
Phase 2: Extract Content
    ↓
Phase 3: Analyze Content
    ↓
Phase 4: Select Output Format
    ↓
Phase 5: Execute Export
    ↓
Phase 6: Verify and Report

🛑 MANDATORY WORKFLOW - Never skip this!

Must be executed in Antigravity/Gemini CLI:

STEP 1: Check user preferences (Phase 1.5) - Required!

Before processing content must ask the user the following questions:

📊 상세 수준: 1.요약 / 2.보통 / 3.상세
🎯 중점 영역: A.개념 / B.실용 / C.기술 / D.인사이트 / E.전체
📝 노트 분할: ①단일 / ②주제별 / ③원자적 / ④3-tier
🔗 연결 수준: 최소 / 보통 / 최대

기본값(3.상세, E.전체, ④3-tier, 최대)을 사용하시겠습니까?

💡 3-tier란? 개요 노트 + 주제별 노트 + 원자적 노트로 계층 구조화

If it is a social media (Threads/Instagram) URL, additional questions:

🔄 답글 수집 범위:
  1) depth=1: 직접 답글만 (빠름)
  2) depth=2: 답글의 답글까지 (더 완전한 맥락)

⚠️ You must not skip this step!

Can be omitted only if the user used quick preset keywords such as "quickly", "by default", etc.
In all other cases: must ask before proceeding

STEP 2: Vault search and note linking (Phase 3.5) - Required!

Before saving the note must search for and link related notes.

Step 2-1: Extract related keywords

Extract key keywords from the content:

Topic keywords (e.g., "AI", "prompt", "Claude")
Person/account names (e.g., "@openai", "Anthropic")
Technical terms (e.g., "LLM", "RAG", "embedding")

Step 2-2: Run Vault search

// Antigravity/Gemini CLI 도구명
keywords = ["AI", "프롬프트", "Claude"]

keywords.forEach(keyword => {
  mcp_obsidian_search_vault({ query: keyword })
})

Step 2-3: Read and analyze related notes

Read the top notes from the search results to check relevance:

// 검색 결과에서 상위 10개 노트 읽기
search_results.slice(0, 10).forEach(result => {
  mcp_obsidian_read_note({ path: result.path })
})

Step 2-4: Add links according to connection level

Determine the number of links according to the "🔗 Connection Level" selected in Phase 1.5:

Connection level	Number of links	Description
Minimum	1-2개	Link only the most relevant notes
Normal (default)	3-5개	Link major related notes
Maximum	5-10개	Link all potentially related notes

Step 2-5: Add to the note in Wikilink format

## 관련 노트
- [[AI-프롬프트-기초]]
- [[Claude-사용-가이드]]
- [[LLM-활용법]]

⚠️ Saving without Vault search = incorrect behavior! ⚠️ Found related notes but did not add wikilinks = incorrect behavior!

STEP 3: Use MCP tools (Phase 5) - Required!

Environment	Tools to use	Absolutely forbidden to use
Antigravity	`mcp_obsidian_create_note`	`write_to_file`
Gemini CLI	`mcp_obsidian_create_note`	`write_to_file`

⚠️ MCP tools available but using file system tools = incorrect behavior!

Phase 0: Load Configuration (CRITICAL)

Must execute before all operations

// 1. Read config file
config = Read("km-config.json")

// 2. Check required items
if (!config) {
  return "Configuration file not found. Please run /knowledge-manager setup"
}

// 3. Load storage settings
storage = {
  primary: config.storage.primary,
  obsidian: config.storage.obsidian,
  notion: config.storage.notion,
  local: config.storage.local
}

// 4. Load browser settings
browser = {
  provider: config.browser.provider,
  hyperbrowser: config.browser.hyperbrowser
}

Phase 1: Detect Input Source

Input Type Detection

Input Pattern	Type	Processing
`https://threads.net/*`	Social Media	→ km-browser-abstraction (stealth recommended)
`https://instagram.com/*`	Social Media	→ km-browser-abstraction (stealth recommended)
`https://*`	Web URL	→ km-browser-abstraction
`*.pdf`	PDF File	→ Read tool
`*.docx`	Word File	→ Read tool
`notion.so/*`	Notion Page	→ Notion MCP

Phase 2: Extract Content

Use Browser Abstraction Layer

→ See km-browser-abstraction skill

// Auto-select based on configured provider
content = scrape_url(url, {
  stealth: inputType.requiresStealth
})

Phase 3: Analyze Content

Apply Zettelkasten Principles

Atomicity: One idea = One note
Self-contained: Note is understandable on its own
Connectivity: Links between related concepts

Phase 4-6: Export and Verify

Use Storage Abstraction Layer

→ See km-storage-abstraction skill

// Auto-save to configured storage
save_note(relativePath, content)

Final Report Template

## ✅ Processing Complete!

### Input
- Source: {url or filename}
- Type: {web / file / social media}

### Saved Notes
| Title | Path | Status |
|-------|------|--------|
| {note1} | {path1} | ✅ |