mind-notion-sync

name	mind-notion-sync
description	Bidirectional sync between Notion Projects & Tasks databases and Obsidian 200 Projects/ vault structure. Use for "sync Notion", "update projects from Notion", "push to Notion", "sync tasks", "pull from Notion", or any Notion-Obsidian project/task synchronization request.
compatibility	[{"tool":"mcp__claude_ai_Notion__notion-fetch"},{"tool":"mcp__claude_ai_Notion__notion-search"},{"tool":"mcp__claude_ai_Notion__notion-update-page"},{"tool":"mcp__claude_ai_Notion__notion-create-pages"}]

Mind - Notion Sync

Bidirectional synchronization between Notion Projects & Tasks databases and the Obsidian 200 Projects/ vault structure.

Core Principle

Every piece of data written to a file must come from an actual Notion API response or an actual Obsidian file read. Never invent, summarize, or fabricate content. If a fetch fails, skip the record and log it.

When To Use

Use this skill when the user asks to:

• sync projects or tasks between Notion and Obsidian
• pull latest from Notion into Obsidian
• push Obsidian changes to Notion
• update projects from Notion
• check what changed in Notion since last sync
• create a new project/task in both systems

This skill is NOT for:

• querying Notion data without writing files (use Notion MCP directly)
• managing the Neurons Lab workspace (this skill targets the personal workspace only)

Reasoning Order

Before syncing, load context in this order:

1. 000 OS/ — principles and strategy
2. 3 Numerical Targets 2026 — current year targets
3. 300 Areas/3 Mind/ — mind area guidance (LMS/KMS context)
4. Notion MCP — fetch live database state
5. 200 Projects/ — current vault state

Workspace Safety

CRITICAL: Two Notion workspaces are connected — personal and Neurons Lab. This skill operates ONLY on the personal workspace. Before any sync operation:

1. Use notion-search to verify you are querying the personal workspace
2. The Projects database lives at collection://<YOUR-PROJECTS-COLLECTION-ID> (discover once via notion-search, then reuse)
3. The Tasks database lives at collection://<YOUR-TASKS-COLLECTION-ID>
4. If results include work-workspace data, filter it out

Notion Database Schema

Projects Database

Notion Property	Type	Obsidian Property	YAML Type
Project name	Title	title + filename	text
Area	Select	area	text
Status	Status	status	text
Priority	Select	priority	text
Owner	Person	owner	text
Deadline	Date	deadline	date
Completion	Rollup	SKIP	—
Tasks	Relation	SKIP (Dataview)	—

Area options: Work, Lifestyle, Society, Mind Status options: Backlog, Paused, Ongoing, In Progress, On Others, Done Priority options: Low, Medium, High

Tasks Database

Notion Property	Type	Obsidian Property	YAML Type
Task name	Title	title + filename	text
Status	Status	status	text
Priority	Select	priority	text
Task type	Select	task-type	text
Assignee	Person	assignee	text
Due	Date	due	date
Project	Relation	project	wikilink
Sub-tasks	Relation	sub-tasks	wikilink list
Parent-task	Relation	parent-task	wikilink
Project Status	Rollup	SKIP	—

Status options: Backlog, Planned, In Progress, On Others, Done, Archived Task type options: Low/Operational, Average/Professional, Creative/Technical, Leverage/Systems

Status Semantics

Both "Done" and "Archived" count as completed. All Dataview queries must use:

• Active filter: status != "Done" AND status != "Archived"
• Completed filter: (status = "Done" OR status = "Archived")
• Progress count: t.status === "Done" || t.status === "Archived"

Obsidian Vault Structure

200 Projects/
├── _Templates/
│   ├── project-template.md
│   └── task-template.md
├── _Bases/
│   ├── Projects.base
│   └── Tasks.base
├── _Dashboard.md
├── _migration/
│   └── notion-export-raw.json
├── {Project Name}/
│   ├── {Project Name}.md
│   └── tasks/
│       └── {Task}.md
├── _Unassigned Tasks/
│   └── {Orphan Task}.md
└── Old projects notes/          ← NEVER TOUCH

Workflow A — Notion → Obsidian (Pull)

Use when the user says "sync from Notion", "pull from Notion", "update from Notion".

Step 1: Fetch all projects

Search the Projects data source for all pages:

mcp__claude_ai_Notion__notion-search(
  query: "projects",
  data_source_url: "collection://<YOUR-PROJECTS-COLLECTION-ID>",
  page_size: 25
)

Do multiple searches with different queries to ensure completeness. Then fetch each project page individually with notion-fetch to get full properties and body content.

Step 2: Fetch all tasks

Same approach for the Tasks data source (collection://<YOUR-TASKS-COLLECTION-ID>). Additionally, collect task IDs from each project's Tasks relation property — this is the authoritative source for project-task associations.

Fetch EVERY task page individually with notion-fetch. Never skip a fetch.

Step 3: Build lookup and resolve relations

Create a map: {notion_page_id → title} for all projects and tasks. Replace every relation URL/ID with the corresponding "[[Title]]" wikilink.

Step 4: Determine sync actions

For each Notion record:

• If a file with matching notion-id exists in the vault → UPDATE the file
• If no matching file exists → CREATE a new file
• If a vault file has a notion-id that no longer exists in Notion → LOG as potentially deleted (do not auto-delete)

Step 5: Write/update project files

For each project, ensure the folder exists at 200 Projects/{SanitizedName}/ with a tasks/ subfolder. Write or update the project .md file with:

---
title: "{project name}"
type: project
area: "{area}"
status: "{status}"
priority: "{priority}"
owner: "{owner}"
deadline: {YYYY-MM-DD or empty}
source: "Notion"
created: "{original created date or today}"
notion-id: "{uuid-with-dashes}"
tags:
  - project
---

# {Project Title}

{Converted Notion body content — VERBATIM, not summarized}

## Tasks

```dataview
TABLE status AS "Status", priority AS "Priority", due AS "Due", assignee AS "Assignee"
FROM "200 Projects"
WHERE type = "task" AND contains(project, this.file.link) AND status != "Done" AND status != "Archived"
SORT due ASC

Completed Tasks

TABLE due AS "Completed"
FROM "200 Projects"
WHERE type = "task" AND contains(project, this.file.link) AND (status = "Done" OR status = "Archived")
SORT due DESC

Progress

let tasks = dv.pages('"200 Projects"')
  .where(t => t.type === "task" && t.project && dv.equal(t.project, dv.current().file.link));
let done = tasks.where(t => t.status === "Done" || t.status === "Archived").length;
let total = tasks.length;
let pct = total > 0 ? Math.round((done/total)*100) : 0;
dv.paragraph(`**${done}/${total} tasks complete (${pct}%)**`);
dv.paragraph(`<progress value="${done}" max="${total}" style="width:100%"></progress>`);

### Step 6: Write/update task files

For each task:
- With project → `200 Projects/{ProjectFolder}/tasks/{SanitizedTaskName}.md`
- Without project → `200 Projects/_Unassigned Tasks/{SanitizedTaskName}.md`

```markdown
---
title: "{task name}"
type: task
status: "{status}"
priority: "{priority}"
task-type: "{task type}"
assignee: "{assignee}"
due: {YYYY-MM-DD or empty}
project: "[[{Project Name}]]"
parent-task: "{wikilink or empty}"
sub-tasks: [{wikilink list}]
source: "Notion"
created: "{date}"
notion-id: "{uuid-with-dashes}"
tags:
  - task
---

# {Task Title}

**Project:** `= this.project`

{Converted Notion body content — VERBATIM}

Step 7: Print sync summary

=== SYNC SUMMARY (Notion → Obsidian) ===
Projects: X synced (Y new, Z updated)
Tasks: A synced (B new, C updated)
Orphan tasks: D (in _Unassigned Tasks/)
Skipped/failed: [list]

Workflow B — Obsidian → Notion (Push)

Use when the user says "push to Notion", "update Notion", "sync to Notion".

Step 1: Scan vault

Read all .md files in 200 Projects/ with type: project or type: task in frontmatter. Skip Old projects notes/, _Templates/, _Dashboard.md.

Step 2: Classify changes

For each file:

• Has notion-id → potential UPDATE to existing Notion page
• No notion-id → potential CREATE in Notion

Step 3: Push updates

For files with notion-id, fetch the current Notion page and compare properties. If the Obsidian frontmatter differs, update the Notion page:

mcp__claude_ai_Notion__notion-update-page(
  page_id: "{notion-id}",
  properties: { mapped property changes }
)

Map Obsidian properties back to Notion format:

• status → Status property
• priority → Select property
• area → Select property
• deadline/due → Date property
• title → Title property

Step 4: Push new records

For files without notion-id, create a new Notion page in the appropriate database:

• type: project → create in Projects database
• type: task → create in Tasks database, set Project relation if project wikilink exists

After creation, update the Obsidian file's notion-id with the new page ID.

Step 5: Print push summary

=== SYNC SUMMARY (Obsidian → Notion) ===
Projects pushed: X (Y new, Z updated)
Tasks pushed: A (B new, C updated)
Skipped: [list with reasons]

Content Conversion Rules

When converting Notion enhanced markdown to Obsidian markdown:

Notion Markup	Obsidian Equivalent
<empty-block/>	Remove
<database ...>Tasks</database>	Remove (replaced by Dataview)
<details><summary>T</summary>C</details>	> [!info]- T + > C (callout)
<columns><column>...</column></columns>	Sequential content with ---
<table> with <tr>/<td>	Markdown table
<span color="...">text</span>	Plain text
~~text~~	Keep (strikethrough)
- [x] / - [ ]	Keep (checkboxes)
	Keep as 
<mention-page url="..."/>	Remove or [[Note Name]] if mappable
Other XML/HTML tags	Remove

Filename Sanitization

• Replace /\:*?"<>| with -
• Collapse multiple spaces or dashes to single
• Trim leading/trailing whitespace and dots
• Maximum 200 characters
• Empty title → Untitled-{first 8 chars of notion-id}
• Collision → append (2), (3) etc.

YAML Safety Rules

• kebab-case property names: task-type, parent-task, sub-tasks, notion-id
• Quote strings containing : # [ ] { } , & * ! | > ' " % @
• Wikilinks must be quoted: project: "[[Name]]"
• Empty strings: ""
• Empty lists: []
• Dates: bare ISO format YYYY-MM-DD

Output Contract

After any sync operation, report:

1. Direction — Pull (Notion→Obsidian) or Push (Obsidian→Notion)
2. Records processed — projects and tasks with counts
3. Actions taken — created, updated, skipped
4. Conflicts — records that differ in both systems
5. Errors — failed fetches or writes with IDs
6. Caveats — expired image URLs, missing relations, etc.

Data Quality Rules

• Confidence: High when all records fetched individually. Medium when using cached/search data. Low when records were skipped.
• Image URLs: Notion CDN URLs are temporary (signed S3). Include them but note they may expire.
• Relations: If a relation target page doesn't exist in the extracted data, log it as an orphan reference.
• Body content: Always fetch each page individually with notion-fetch to get the full body. Never rely on search result snippets.

Escalation Rules

• If the user asks about project metrics or status summaries without requesting a sync → use Notion MCP directly, don't run the full sync
• If the user asks about habits or streaks → delegate to mind:streaks-export-analysis
• If the user asks about body/health data → delegate to body:body-data-qa

Tone

Operational and precise. Report counts and actions taken. Flag conflicts clearly. Do not editorialize about project priorities or task importance — just sync the data faithfully.