Tags let you attach structured metadata to documents so the Knowledge block can filter results precisely — by department, date, priority, status, or any dimension you define.
How Tags Work
Tags have two layers:
- Tag definitions — created at the knowledge base level. A definition has a name (e.g., "Department") and a type (Text, Number, Date, or Boolean). Definitions are shared across all documents.
- Tag values — set per document. Once a definition exists, you assign a value to it on each document that needs it (e.g.,
Department = "engineering").
Tag Slots
Each knowledge base has 17 tag slots distributed across four types:
| Type | Slots | Accepted values |
|---|---|---|
| Text | 7 | Any string — matching is case-insensitive |
| Number | 5 | Any valid number |
| Date | 2 | YYYY-MM-DD format |
| Boolean | 3 | true or false |
The type dropdown in the creation form shows current slot usage for each type (e.g., Text (0/7) means none of the 7 text slots are in use yet).
Slots are shared across all documents and connectors in a knowledge base. Connectors that auto-populate metadata tags draw from the same pool. Plan your schema with that in mind.
Defining Tags
Tag definitions live at the knowledge base level. To manage them, click the knowledge base name in the header to open the context menu and select Tags:
This opens the Tags modal, which lists all defined tags and shows how many documents each one is assigned to. Click Add Tag to define a new one:
Enter a Tag Name and pick a Type, then click Create Tag. The name must be unique within the knowledge base. The type dropdown only shows types that still have available slots. Press Enter to submit or Escape to cancel.
To delete a tag definition, click the trash icon next to it. Deleting a definition removes the tag value from every document it was assigned to — the modal shows you which documents are affected before you confirm.
Clicking any existing tag definition opens a dialog showing all documents that have a value set for it, along with their current tag values.
Setting Tag Values on Documents
Once a definition exists, you assign values document by document. Right-click any document (or click the … menu) to open the document context menu, then select Tags:
This opens the tag panel for that document where you can set a value for each defined tag.
Viewing Tags in the Document List
The Tags column in the document list shows the current tag values for each document at a glance. Documents with no tags assigned show – – –:
Use the Filter and Sort controls in the top right to narrow the list by tag values or sort by them.
Using Tags in the Knowledge Block
In a workflow, open the Knowledge block and configure Tag Filters to restrict which documents are searched:
Each filter has three parts:
- Tag — select a tag definition from the knowledge base
- Operator — depends on the tag type (see below)
- Value — the value to match against
Add as many filters as you need with the + button. Multiple filters are combined with AND logic — a document must match all filters to be included in the search.
Operators by Type
| Type | Available operators |
|---|---|
| Text | equals, not equals, contains, does not contain, starts with, ends with |
| Number | equals, not equals, greater than, greater than or equal, less than, less than or equal, between |
| Date | equals, after, on or after, before, on or before, between |
| Boolean | is, is not |
Tag values in filter fields can be static strings or workflow variable references (e.g., <start.department>), so filtering can adapt dynamically at runtime.
Search Modes
The Knowledge block behaves differently depending on what you provide:
| What you provide | Behaviour |
|---|---|
| Tags only (no search query) | Fetches all documents that match the tag filters — pure tag matching, no vector search |
| Query only (no tag filters) | Semantic vector search across all documents in the knowledge base |
| Both tags and query | Tag filters run first to narrow the document set, then vector search runs within that subset |
The combined mode is the most precise — tag filtering cuts down the candidate set cheaply before the more expensive vector similarity comparison runs.
Connector-Populated Tags
Connectors can auto-populate tags with metadata from the source. A Notion connector might set Last Modified and Labels; a GitHub connector might set Repository and File Path. These work exactly like manually defined tags and are available in Knowledge block filters.
You can disable specific metadata tag types during connector setup or in connector settings to free up slots for manual use. See Connectors for details.