Once your data is indexed, the Knowledge tab becomes your control center. See at a glance which documents are searchable, which are still processing, and which need attention — then organize, reindex, or remove content to keep your search results accurate.
Document list
The document list is organized with filter tabs across the top and a table below.
Filter tabs
Quickly find what you’re looking for by filtering documents by source:
| Tab | What you’ll see |
|---|
| All | Every document regardless of source |
| Folders | Documents organized into folders within this knowledge base |
| Files | Documents added via file upload |
| Integrations | Documents synced from connected apps |
| URLs | Documents added via URL scraping |
| Tables | Documents indexed from VectorShift tables |
Columns
| Column | What it shows |
|---|
| Name | The document name (auto-generated from the file name and timestamp) |
| Type | The file type (e.g., Plain Text Document) |
| Created Date | When the document was added |
| Status | Current processing status |
Status values
Know exactly where each document stands in the indexing process:
| Status | What it means |
|---|
| Success | Ready to search — fully processed and indexed |
| Processing | In progress — being chunked, embedded, and indexed (shown as a spinner) |
| Failed | Something went wrong — click Retry to try again |
| Warning | Partial issues — shown for folders when one or more items inside failed to index |
Alert banners
In certain situations, the document list displays informational alert banners:
- “Parent folder already selected to Sync”: A parent folder is already queued for syncing. To customize indexing settings for individual items inside, remove the parent folder sync first.
- “Items cannot be deleted individually”: Documents inside a recursive URL folder can only be removed by deleting the entire folder.
Other list controls
- Search: Find a document by name using the search bar.
- + Add folder: Create a folder within the knowledge base to organize your documents. Folder names must be at least 3 characters long and can only contain letters, numbers, and underscores.
- Share: Share documents with other users.
- + Add Document: Add more content directly — opens a dropdown with all available indexing options (upload files, add integration, scrape URL, etc.).
Bulk actions
Save time by acting on multiple documents at once. Select documents using the checkboxes, then use the bulk action buttons:
| Action | What it does |
|---|
| Reindex | Re-processes all selected documents with the current indexing settings |
| Move | Moves all selected documents to a folder within the knowledge base |
| Delete | Permanently removes all selected documents |
Actions on individual documents
Each document row has action icons on the right side and a three-dot menu.
Need different processing for a specific document? Click the settings icon (first icon) to override the knowledge base defaults and reindex with custom settings — without affecting any other documents.
The dialog has two sections:
Permanent settings
These reflect the knowledge base-level settings. Some can be toggled per document:
- Embedding model: Shows the model selected during knowledge base creation (read-only).
- Advanced document analysis (beta): Toggle on or off for this document.
- Hybrid search: Toggle on or off for this document.
Indexing settings
Customize these for this specific document to improve its search results:
- Chunk size: Adjust the slider or enter a value. Overrides the knowledge base default for this document only.
- Chunk overlap: Adjust the slider or enter a value. Must be less than the chunk size.
- Splitter method: Choose Sentence, Markdown, or Dynamic.
- Processing model: Choose the processing model for this document.
Click Re-index to reprocess the document with the new settings, or Cancel to discard.
This is especially useful when one file type needs different handling — for example, a scanned PDF that needs Mistral OCR while your other docs work fine with the Default model.
Item preview
Verify how VectorShift processed your document — inspect chunks, parsed content, and metadata to make sure everything looks right. Open the preview by:
- Clicking the preview icon (second icon) on the document row
- Clicking directly on the document name in the list
Both open the same preview drawer with three tabs:
| Tab | How it helps |
|---|
| Chunks | See how your content was divided — useful for diagnosing why search results may be too broad or too narrow |
| Parsed | View the extracted text as the processing model sees it — verify nothing was missed or misread |
| Metadata | Check the document’s metadata, including any auto-generated fields — confirm tags are accurate |
Click the three-dot icon (last icon) to access additional options:
| Option | What you can do |
|---|
| Copy ID | Copy the document’s unique identifier to your clipboard — useful for API calls or pipeline configurations |
| Move | Organize by moving the document to a folder within this knowledge base |
| Delete | Permanently remove the document from the knowledge base |
