notion-ingest

DEPRECATED: This plugin has been superseded by ai-zettelkasten. Please migrate to ai-zettelkasten for continued updates. This plugin is no longer maintained.

Notion Document Ingestion

When user provides a Notion URL or page ID, or says "ingest from notion", follow this process to atomize the document into Zettelkasten notes.

Trigger Phrases

• /notion-ingest <url or page_id>
• "ingest notion page"
• "import from notion"
• "add notion doc to zettelkasten"

Dual-Storage Architecture

Notes are stored in two places:

Storage	Purpose	Tool
ChromaDB	Vector search, semantic queries, metadata filtering	mcp__chroma__chroma_add_documents
Obsidian	Human-readable markdown, wikilinks, graph view	mcp__obsidian__write_note

Both must be updated together to keep the systems in sync.

Step 1: Fetch Content

Extract page ID from URL if needed. Notion URLs have format: https://www.notion.so/Page-Title-{page_id}

The page_id is the 32-character hex string at the end (may have hyphens).

# Get page metadata
mcp__MCP_DOCKER__API-retrieve-a-page(page_id)

# Get page content blocks
mcp__MCP_DOCKER__API-get-block-children(block_id=page_id, page_size=100)

Parse the block children to extract:

• Headings (heading_1, heading_2, heading_3) → Section structure
• Paragraphs → Body content
• Bulleted/numbered lists → Supporting points
• To-do blocks → Potential tasks
• Quotes → Key citations
• Links → Source references

Step 2: Analyze Structure

Count and categorize:

Block Types:
- Headings: {count} (section boundaries)
- Paragraphs: {count}
- Lists: {count}
- TODOs: {count} (potential tasks)
- Quotes: {count}
- Links: {count} (potential sources)

Identify:

1. Main sections by heading_2 blocks (these become atomic notes)
2. Key concepts that deserve their own notes
3. Research items from TODO blocks
4. Sources from URLs/citations
5. Personal reflections vs external content

Step 3: Propose Atomization

Present to user before creating:

Proposed Structure for: "{document_title}"
Source: {notion_url}

INDEX NOTE:
  ID: {topic}-1
  Title: {document_title}
  Type: index-note

ATOMIC NOTES ({count}):
  {topic}-1a: {section_1_title} (literature-note)
  {topic}-1b: {section_2_title} (literature-note)
  {topic}-1c: {section_3_title} (permanent-note)
  ...

RESEARCH TASKS ({count}):
  task-{topic}-1: {todo_1}
  task-{topic}-2: {todo_2}
  ...

LINKS:
  {topic}-1 (parent)
  ├── {topic}-1a (child) ←→ task-{topic}-1 (related)
  ├── {topic}-1b (child)
  └── {topic}-1c (child) ←→ {topic}-1a (related)

Proceed? [y/n] or suggest changes:

Step 4: ID Generation

Follow Luhmann-style branching:

• Ask: "What topic prefix should we use?" (e.g., llm, design, learn)
• Check existing IDs with that prefix
• Generate: {prefix}-{next_number} for index
• Children: {prefix}-{number}a, {prefix}-{number}b, etc.
• Tasks: task-{topic}-{number}

Step 5: Create Notes in ChromaDB

After user approval, create all documents in ChromaDB for vector search:

# For each note, create with chroma_add_documents:
{
  "id": "{generated_id}",
  "document": "{title}\n\n{body}\n\nTags: {tags}\nSource: {source}",
  "metadata": {
    "title": "{title}",
    "category": "{category}",  # index-note, literature-note, permanent-note, fleeting-note, kanban-task
    "topic": "{topic}",
    "tags": "{comma,separated,tags}",
    "source": "{source_citation}",
    "sourceUrl": "{url}",
    "dateAdded": "{ISO_date}",
    "parentIds": "{comma,separated,ids}",
    "childIds": "{comma,separated,ids}",
    "relatedIds": "{comma,separated,ids}"
  }
}

Step 5b: Create Notes in Obsidian

IMPORTANT: Also create each note in Obsidian for browsable markdown with wikilinks.

Obsidian Vault: /Users/rc/Documents/Obsidian Vault Folder: Zettelkasten/

Use mcp__obsidian__write_note for each note:

mcp__obsidian__write_note(
  path="Zettelkasten/{id} {title}.md",
  content="""# {title}

{body_as_markdown}

## Connections

- Parent: [[{parent_id} {parent_title}]]
- Related: [[{related_id} {related_title}]], ...
""",
  frontmatter={
    "id": "{id}",
    "title": "{title}",
    "category": "{category}",
    "topic": "{topic}",
    "tags": ["{tag1}", "{tag2}"],  # Array format for Obsidian
    "source": "{source}",
    "sourceUrl": "{url}",
    "dateAdded": "{ISO_date}",
    "parentIds": ["{parent_id}"],
    "status": "todo"  # For tasks only
  }
)

Obsidian-specific formatting:

• Use [[note-name]] wikilinks for connections (Obsidian resolves these)
• Include filename pattern: {id} {title}.md for readability
• Use YAML arrays for tags (not comma-separated strings)
• Add status field for kanban tasks

Step 6: Establish Links

Ensure bidirectional links:

1. Index note's childIds contains all atomic note IDs
2. Each atomic note's parentIds contains index note ID
3. Tasks' relatedIds link to relevant concept notes
4. Related concepts link to each other via relatedIds

Step 7: Connection Discovery

After creating notes, run semantic search to find connections to existing notes:

chroma_query_documents(
  collection_name="zettelkasten",
  query_texts=["{new_note_content}"],
  n_results=5
)

For each new note, suggest connections:

Potential connections for {id}:
[1] {existing_id}: "{title}" (similarity: 0.XX)
[2] {existing_id}: "{title}" (similarity: 0.XX)

Create links? [1,2] or [n]one:

Category Assignment Guidelines

Content Type	Category
Overview/TOC of a topic	index-note
Summary of external source	literature-note
Your own developed ideas	permanent-note
Quick thoughts, unprocessed	fleeting-note
Action items from TODOs	kanban-task
Large initiatives	epic

Tag Extraction

Auto-extract tags from:

• Explicit tags in document
• Key terms (capitalized concepts, technical terms)
• Topic area
• Status indicators (draft, review, etc.)

Default tags by category:

• literature-note: source:{author}, year:{year}
• kanban-task: todo (status tag)
• fleeting-note: unprocessed

Example Session

User: /notion-ingest https://notion.so/My-Notes-abc123def456