> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cognee.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Obsidian

> Import an Obsidian vault, with its links and tags, as a new dataset

The **Obsidian** card imports a vault from your computer. It is an import, not a connection: there is no sign-in and no daily sync, and each import creates a new dataset with its own knowledge graph. Any workspace member can import, not only the owner.

## Importing a vault

**1. Open the card**

In [Integrations](https://docs.cognee.ai/cognee-cloud/ui/integrations.md) → **Data sources**, click **Import vault** on **Obsidian**.

**2. Drop your vault folder**

Drop the vault folder onto the dialog, or browse to it. Drop one folder only. The dialog shows the limits for your workspace: the number of notes, the size per note, and the total size.

Only `.md` notes are read. Hidden folders such as `.obsidian`, `.git` and `.trash` are skipped.

**3. Review what Cognee found**

Before anything is created, the dialog shows the **Vault name** (editable), the number of **Notes**, the **Size**, the number of **Links** and **Tags**, and your most used tags. It warns about links that point at a note missing or ambiguous in the folder.

**4. Import**

Click **Import \{n} notes**. Cognee uploads the notes, then builds the knowledge graph. You can close the window: the import keeps running. When it finishes, the dialog shows **Graph ready** and the dataset name.

Importing builds a knowledge graph, which uses credits like any other ingestion. Wikilinks and Markdown links between notes become relations in the graph.

> [!NOTE]
>
> If the card reads *Obsidian import is not available on this workspace yet.*, your workspace has not been updated for imports yet. Try again later.

## Importing again

Every import creates a new dataset, so re-importing an updated vault does not replace the previous one. Delete the old dataset on the [Datasets](https://docs.cognee.ai/cognee-cloud/ui/datasets.md) page if you no longer need it.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.