Zettelkasten — AI-Powered Note Organizer for Obsidian
Automatically organize your Obsidian vault using the Zettelkasten method. Drop files into inbox, AI atomizes content into permanent atomic notes, builds wikilinks, and maintains Maps of Content (MOC) — powered by Claude Code.
Zero dependencies. No Python, no scripts — pure Claude Code skills + agents.
中文版
The Problem
You clip articles, jot ideas, save snippets — and they pile up in a folder, unlinked and forgotten. Manual note organization doesn't scale. Your second brain stays dumb.
The Solution
This plugin turns your Obsidian vault into a self-organizing knowledge base:
- Atomize — One concept per note. Multi-topic files are automatically split.
- Rewrite — Clean prose with proper structure, preserving all original information.
- Link — Every note connects to existing knowledge via contextual wikilinks. Cross-domain connections preferred.
- Navigate — Maps of Content (MOC) auto-generated when topics accumulate ≥3 notes.
Quick Start
1. Install the Plugin
# In Claude Code
/plugins install github:henrywen98/zettelkasten
2. Set Up Your Vault
cd /path/to/your/obsidian/vault
mkdir -p 0_inbox 1_zettel 2_maps 3_output 4_assets
git init # recommended for version safety
3. Ingest Your First Notes
Drop any files (notes, web clips, article exports) into 0_inbox/, then:
cd /path/to/your/vault
claude
> /zet-ingest
That's it. Atomic notes appear in 1_zettel/, organized by year-month, interlinked, with MOC navigation in 2_maps/.
Features
/zet-ingest — Inbox to Knowledge Base
The core workflow. Scans 0_inbox/, dispatches AI agents to process files in batches of ~10, then updates MOCs and commits via git.
Each file goes through: read → classify → atomize → rewrite → frontmatter → link → write.
0_inbox/weekly-learning.md → 1_zettel/2026-04/ssh-key-authentication.md
(SSH + Python decorators + Docker) 1_zettel/2026-04/python-decorator-patterns.md
1_zettel/2026-04/docker-compose-networking.md
- Multi-topic files split into independent atomic notes
- Each note links to ≥1 existing note (connection forcing)
- Original language preserved (Chinese stays Chinese, English stays English)
- Source files deleted after processing — inbox stays clean
/zet-query — Ask Your Knowledge Base
Query your accumulated knowledge in natural language. The AI navigates MOCs and note links, reads relevant notes, and synthesizes an answer with [[wikilink]] citations.
> /zet-query What do I know about authentication?
Output saved to 3_output/ with full source references.
/zet-lint — Vault Health Check
Structural integrity scanner:
- Broken wikilinks pointing to non-existent notes
- Orphan notes with no inbound or outbound links
- Incomplete frontmatter (missing required fields)
- MOC coverage gaps — uncategorized notes
- Stale MOC counts
Auto-fix offered for common issues.
Vault Structure
Vault/
├── 0_inbox/ Drop materials here (deleted after processing)
├── 1_zettel/ Permanent notes — atomic, linked, by year-month
│ └── YYYY-MM/ e.g. 2026-04/
├── 2_maps/ Maps of Content — auto-maintained topic navigation
├── 3_output/ Query results, lint reports
└── 4_assets/ Images and attachments
Note Format
Every permanent note follows a consistent structure with YAML frontmatter:
---
id: "202604061430"
title: "SSH Key Authentication"
created: 2026-04-06
processed: 2026-04-06
source: web-clip # original | web-clip | import
tags: [ssh, security, devops]
summary: "How SSH key-based authentication works — key generation, exchange, and verification flow"
---
# SSH Key Authentication
[Note content in clear prose...]
## Links
- Related to [[tls-handshake-protocol]]: both use asymmetric cryptography for initial authentication
- See [[server-hardening-checklist]]: SSH key auth is a key step in hardening
How It Works
/zet-ingest
│
▼
zet-ingest (skill) ← Orchestrator: scan, batch, dispatch, MOC, commit
│
├── zet-worker (agent) ← Batch 1: read → atomize → rewrite → link → write
├── zet-worker (agent) ← Batch 2: can link to batch 1's notes
└── ... ← Sequential processing, each batch commits before next
