← Back to docs

Connect Obsidian to your AI tools with Vaulken and S3

Your Obsidian vault is full of valuable notes — meeting notes, project docs, research, ideas. What if your AI assistant could read, search, and reference them directly?

With Vaulken and an S3 sync plugin, you can connect your Obsidian vault to any MCP-compatible AI tool (Claude Desktop, Cursor, Windsurf) in three steps.

What you'll need

  • An Obsidian vault with notes you want to make accessible to AI
  • An S3-compatible storage bucket (AWS S3, Scaleway Object Storage, MinIO, Cloudflare R2)
  • A Vaulken instance (cloud or self-hosted)

Step 1: Sync Obsidian to S3

The Remotely Save plugin syncs your Obsidian vault to S3-compatible storage.

Install the plugin

  1. Open Obsidian → Settings → Community Plugins → Browse
  2. Search for Remotely Save
  3. Install and enable

Configure S3

  1. Go to Remotely Save settings
  2. Select S3 or S3-compatible as the remote type
  3. Fill in your S3 credentials:
    • Endpoint — your S3 endpoint (e.g., s3.fr-par.scw.cloud for Scaleway)
    • Region — your bucket region (e.g., fr-par)
    • Bucket — your bucket name
    • Access Key ID — your S3 access key
    • Secret Access Key — your S3 secret key
  4. Optionally set a base directory if you want your notes in a subdirectory (e.g., obsidian/)
  5. Click Check connectivity to verify
  6. Run the initial sync

Your notes are now in S3.

Step 2: Create a Vaulken vault

Now point Vaulken at the same S3 bucket.

  1. Log in to your Vaulken dashboard
  2. Click Create vault
  3. Select Bring your own bucket (BYOB)
  4. Enter the same S3 connection details: endpoint, region, bucket name, access key, secret key
  5. Save

Vaulken now has access to your files. Full-text search is available out of the box. Semantic search requires enabling the background indexing service (see your instance's documentation).

Step 3: Connect your AI tool

Create an Access Config to let your MCP client access the vault.

  1. Open your vault in the Vaulken dashboard
  2. Go to Access Configs → Create
  3. Set permissions:
    • Read — your AI can read and search your notes
    • Read + Write — your AI can also create or edit notes
  4. Optionally set a path prefix to restrict access to a subdirectory (e.g., obsidian/)
  5. Save and copy the MCP server URL displayed on the Access Config page

After saving, copy the MCP server URL shown on the Access Config page. It looks like:

https://your-vaulken-instance.com/mcp/{accessId}

Vaulken works with any MCP-compatible client. Here's how to set it up with the most popular ones.

Claude.ai

  1. Go to claude.ai → Settings → Integrations
  2. Click Add integration
  3. Paste your MCP server URL
  4. Authorize access when prompted in the browser popup
  5. Your Obsidian notes are now available in your Claude conversations

Claude Desktop

Add this to your Claude Desktop MCP configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "obsidian-vault": {
      "url": "https://your-vaulken-instance.com/mcp/your-access-id"
    }
  }
}

On first connection, Claude Desktop discovers the authorization server automatically and prompts you to authorize in your browser.

Claude Code

claude mcp add obsidian-vault https://your-vaulken-instance.com/mcp/your-access-id

Claude Code handles the OAuth flow automatically on first use.

Cursor / Windsurf

Go to Settings → MCP Servers → Add and paste the same MCP server URL. Both editors handle the authorization flow the same way.

What you can do now

Once connected, your AI assistant can:

  • Search your notes — "Find all my notes about project X"
  • Read specific files — "Read my meeting notes from last week"
  • Cross-reference — "Summarize what I know about topic Y across all my notes"
  • Create new notes — "Write a summary of our conversation and save it to my vault"
  • Semantic search — "Find notes related to this concept" (uses meaning, not just keywords)

Tips

  • Sync frequency: Remotely Save can sync on startup, on interval, or manually. For near real-time access, set a short interval (5 minutes).
  • Path prefix: If your S3 bucket contains other data, use a path prefix like obsidian/ in both Remotely Save and your Vaulken Access Config.
  • Start read-only: Begin with read-only access. Enable write access once you're comfortable, and consider a separate directory for AI-generated content.
  • Multiple tools: Create separate Access Configs with different path prefixes for different AI tools — each gets its own scope.

Going further

  • Enable semantic search to let your AI find notes by meaning, not just keywords
  • Use Vaulken's frontmatter tools to manage Obsidian properties (tags, dates, status) programmatically
  • Set up a shared team vault where multiple people sync notes and AI tools can search across everyone's knowledge