obsidian-import

Obsidian Import Skill

Import an Obsidian vault into ChromaDB for semantic search. Parses markdown structure (frontmatter, headings, tags, links, tables) and chunks content by heading hierarchy so each vector represents a coherent topic.

Requirements

pip install chromadb pyyaml

Quick Start

# Scan a vault (dry run — no writes)
python3 {baseDir}/obsidian_import.py scan --vault ~/my-vault

# Import everything into ChromaDB
python3 {baseDir}/obsidian_import.py import --vault ~/my-vault

# Import only files in a specific folder
python3 {baseDir}/obsidian_import.py import --vault ~/my-vault --folder "Projects/Active"

# Import only files with a specific tag
python3 {baseDir}/obsidian_import.py import --vault ~/my-vault --tag "important"

# Incremental import (skip unchanged files)
python3 {baseDir}/obsidian_import.py import --vault ~/my-vault --incremental

Commands

scan — Preview what would be imported

python3 {baseDir}/obsidian_import.py scan --vault ./my-vault [--folder PATH] [--tag TAG] [--glob PATTERN]

Shows file count, section count, total chunks, and per-file breakdown. No writes — safe to run anytime.

import — Import vault content into ChromaDB

python3 {baseDir}/obsidian_import.py import --vault ./my-vault \
  [--collection vault] \
  [--folder PATH] \
  [--tag TAG] \
  [--glob PATTERN] \
  [--incremental] \
  [--chunk-size 1500] \
  [--chunk-overlap 150]

Parses every matching .md file, splits by heading sections, chunks long sections, and upserts into ChromaDB with rich metadata.

clear — Remove all items from a collection

python3 {baseDir}/obsidian_import.py clear --collection vault

Deletes the collection entirely. Use before a full re-import if needed.

stats — Show collection statistics

python3 {baseDir}/obsidian_import.py stats [--collection vault]

How It Works

1. File Discovery

Recursively finds all .md files in the vault, excluding:

• .obsidian/ (config directory)
• .trash/ (Obsidian trash)
• Files matching --glob exclusion patterns
• Dot-prefixed directories