Skip to main content
Bring your Obsidian vault into Membase as a set of wiki documents in one drag-and-drop. Folder structure is flattened into collections, [[wikilinks]] are preserved, and you decide whether to keep the original markdown or let Membase summarize each note.
Obsidian vault importing into Membase
This page covers the import flow itself. If you want to know what you can do with your documents once they are inside Membase (editing, graph view, search, MCP tools), see the Knowledge Wiki guide.

​
How It Works

Once you drop a vault .zip onto the Wiki page, Membase:
  1. Unzips the archive and walks every .md file
  2. Creates one collection per top-level folder
  3. Turns each markdown file into a wiki document
  4. Resolves [[wikilinks]] into real backlinks between documents
  5. Optionally runs a context-aware LLM summarization on each document
Processing runs asynchronously, so you can close the tab and come back when it’s done. You’ll also get an email once the import finishes.

​
Importing a Vault

Export your vault as a .zip

In your file manager (or from within Obsidian), zip your vault folder. Membase accepts the raw vault structure, so you do not need to flatten or rename anything before zipping.The maximum zip size is 5 GB with up to 10,000 files per import. Nested .zip files inside the archive are also supported.

Drop the zip on the Wiki page

Go to the Wiki tab in your dashboard and drag the .zip onto the page (or use the Import Documents button). Uploads use resumable TUS, so large vaults are safe to pause and resume.
Obsidian vault import drop zone

Choose original or summarized content

Toggle LLM Summarization in the import dialog before starting:
  • Off (default): The original markdown is kept exactly as written. Frontmatter is preserved in the document metadata and wikilinks are kept as-is.
  • On: Each document is rewritten by an LLM into a structured, deduplicated summary using your other documents as context. Useful when your notes are long, repetitive, or written for yourself rather than for an agent.
You can mix strategies by running separate imports: bring in some folders as originals and others as summaries.

Let Membase process the import

Imports run in the background via Cloud Tasks. When it finishes, you’ll see a summary of documents created, duplicates skipped, collections created, and backlinks resolved. An email goes out to your account address as well.

​
Folder Structure Is Flattened

Obsidian vaults often nest folders several levels deep. Membase collections are a flat, single-depth concept, so imports are flattened like this:
Obsidian path (after unzipping)Membase location
Engineering/api.mdCollection: Engineering, document: api
Engineering/Auth/oauth.mdCollection: Engineering, document: oauth
Engineering/Auth/OIDC/flows.mdCollection: Engineering, document: flows
notes.md (at archive root)No collection (Basic Collection), document: notes
Only the first folder level becomes a collection. Everything deeper is still imported, but those documents are placed directly in the top-level collection without any sub-folder hierarchy. Markdown files at the archive root land in the default Basic Collection.
If your zip has a single wrapper folder (e.g. vault/Engineering/...), Membase detects it and treats the wrapper as the archive root. If your wrapper folder only contains root-level .md files and no subfolders, those files will be grouped into a collection named after the wrapper. Zip the contents of your vault (not the vault folder itself) to avoid this.
If you rely on nested folders for grouping, flatten them in Obsidian first, or split your vault into multiple zips so each zip represents one collection.
Inline images and attachments are not imported yet. Only markdown text is brought in. Images referenced from your vault (![[image.png]] or relative paths) won’t render until inline image uploads ship. External image URLs in markdown will still render as normal.
Obsidian and Membase use the same model: [[wikilink]] is the forward reference syntax, and backlinks are the reverse-direction view of those references. Both are preserved on import.
  • [[Note Title]] references are kept verbatim in the document content, just like in Obsidian
  • Each reference also creates a backlink edge between the source and target documents
  • Those edges power the Wiki Graph view and the backlinks panel shown on each document page, so you can navigate in both directions
If a link points to a document that does not exist in your vault yet, it stays as plain [[wikilink]] text. It automatically becomes a real backlink the moment you create or import a document with that title.
Wiki document showing wikilinks and backlinks between documents

​
Deduplication

Membase dedupes wiki documents by normalized title across your entire wiki (not per collection). Two notes with the same title cannot coexist, even if they live in different collections.
  • Re-importing the same vault does not create duplicates. Matching documents are skipped, not overwritten.
  • Documents with the same title but in different collections will also be skipped during import, so only one copy exists per user.
  • The import result dialog reports skipped counts separately from errors.
  • To refresh a document, either edit it in the dashboard or delete the old one before re-importing.

​
Limits

LimitValue
Max zip size5 GB
Max files per import10,000
Document quota200 (Free) / 2,000 (Pro)
When your wiki document quota is reached mid-import, remaining files are skipped and the dialog surfaces a partial-completion warning. Already-imported documents stay in your wiki.

​
Using Your Vault After Import

Once the import completes, your notes are regular wiki documents. Everything in the Knowledge Wiki guide applies:
  • Editing: Open any document in the dashboard and edit inline with [[wikilink]] autocomplete
  • Graph view: See your vault as a force-directed network, colored per collection
  • Table view: Sort, filter, and bulk-delete documents
  • Hybrid search: Query by keyword or meaning, scoped to a collection if you like
  • MCP tools: Your connected agents can call search_wiki, add_wiki, update_wiki, and delete_wiki

Knowledge Wiki

Full guide to editing, graph view, search, and MCP tools.

Membase MCP

Parameters for the wiki MCP tools your agents can call.

Bring Your Context

Import chat history and connect more sources.

Use Your Context

Chat with Memory, agent retrieval, and dashboard exploration.