Knowledge Wiki

The Knowledge Wiki is your structured knowledge base inside Membase. While memories capture personal context (preferences, decisions, habits), the wiki stores stable, factual knowledge: documentation, references, notes, specs, and anything else you want to recall by title instead of by conversation.

​

Memory vs Wiki

Membase keeps two complementary knowledge stores. Agents can search both, but each is optimized for a different shape of information.

​

Collections

Collections are folders for grouping related documents. They live in the sidebar where you can create, rename, or delete them. Each collection is also auto-assigned a color that only shows up in the graph view (not in the sidebar or table view), so related documents visually cluster together on the graph. The main way to use collections today is filtering: click a collection in the sidebar to scope the document list, table view, and search results to just that collection. There is no richer per-collection configuration (for example, custom colors or icons) yet.

​

Documents

Wiki documents are markdown files with an extra twist: you can link between them using [[wikilinks]].

• Markdown editor: Headings, lists, tables, code blocks, and other standard markdown formatting
• [[Wikilinks]]: Type [[ in the editor to get instant suggestions from your existing document titles. Click through any link to jump to the referenced document.
• Linked Mentions: Every document shows a Linked Mentions panel with the other documents that reference it (with snippets around each mention), so you can navigate in both directions.
• Manual save: Changes are saved when you click Save or press ⌘/Ctrl + S. There is no autosave.
• Unsaved changes warning: If you try to leave or close the tab with pending edits, Membase warns you before discarding them.

​

Graph View

The wiki graph view visualizes your documents as a force-directed network. Each node is a document, each edge is a [[wikilink]]. Nodes are colored by collection, so you can see at a glance how different areas of knowledge connect.

• Force layout: Drag nodes around, zoom in and out, and watch related documents cluster
• Per-collection colors: Each collection has its own node color for easy scanning
• Click a node: Open the document in the side panel

​

Table View

The table view is a structured list of every document, with sorting, filtering, and pagination.

• Sort by title, collection, updated date, or created date
• Filter by collection or search query
• Pagination for large vaults
• Bulk selection: Select multiple rows to delete in one action

​

Importing an Obsidian Vault

You can bring an entire Obsidian vault (or any folder of markdown files) into the wiki in one drag-and-drop. Top-level folders become collections, [[wikilinks]] are preserved as backlinks, and you can choose whether to keep the original markdown or let Membase summarize each document. See the Obsidian connector guide for the full import flow, folder flattening behavior, size limits, and deduplication rules.

​

Search

The wiki uses hybrid search that combines full-text keyword matching (BM25-style) with semantic similarity and fuses the results with Reciprocal Rank Fusion (RRF). This means you can search either by exact words ("OAuth PKCE") or by meaning ("how do we authenticate users") and get relevant results either way. Search is available in three places:

• Chat with Memory: Ask a question and the agent pulls from both memory and wiki
• Wiki page: The search box filters documents in real time
• Agents via MCP: Your connected agent can call search_wiki during any conversation

​

Scoping search to a collection

You can restrict search to a single collection:

• Dashboard: Click a collection in the sidebar to filter the document list, graph, and search results to that collection only.
• MCP: Pass an optional collection_id to search_wiki to scope results to one collection. Note that this takes a UUID, not a name, so agents typically rely on the user-facing dashboard filter or a user-provided ID rather than discovering collections on their own.

​

Managing Wiki Documents

Unlike memories, wiki documents are fully editable from both the dashboard and your agents.

​

Create

• Dashboard: Click + New Document on the Wiki page, pick a collection (or leave it in the default), and start writing in markdown.
• Obsidian import: Drag a vault .zip onto the Wiki page to bulk-create documents. See the Obsidian connector guide.
• Agent: Your connected agent can call add_wiki to create a document from any conversation.

​

Edit

• Dashboard: Open any document and edit it inline. Click Save (or press ⌘/Ctrl + S) to commit changes. Membase warns you before discarding unsaved edits.
• Agent: Your agent can call update_wiki with a document ID to change the title, content, or collection.

​

Delete

• Dashboard: Delete a single document from the document view, or select multiple rows in the Table View and use the bulk-delete action.
• Agent: Your agent can call delete_wiki with a document ID. Agents should ask for confirmation before deleting.

​

MCP Tools

Your connected agents can read and write wiki documents using these MCP tools. See the Membase MCP reference for full parameters.

​

Document Quotas

Wiki documents are counted against your plan. The current usage is shown on the Billing tab and at the top of the Wiki page. When you reach the limit, creating or importing new documents returns a quota error until you delete existing documents or upgrade your plan.

​

Next Steps