Data types
A Data type defines the structure of a Knowledge record: its fields, how values are validated, which fields identify a record, and which fields nara searches. Every record in Knowledge follows exactly one Data type. Tools use Data types for their arguments and results too.
The Data types page
Section titled “The Data types page”Turn on expert mode, open Building blocks under Automation in the admin console, and select the Data types tab (/admin/knowledge-types). The list shows each type with its name, Usage, whether nara manages it, its version, and its status. nara maintains the managed types; you add your own next to them.
New data type opens Studio. Describe the record you need, review the proposed fields, and publish. Open an existing type to change it the same way.
What a Data type defines
Section titled “What a Data type defines”| Part | Meaning |
|---|---|
| Fields | A JSON schema with strings, numbers, integers, booleans, enums, objects, and references. |
| References | A field that links to records of another Data type. It follows the latest version or a pinned one. |
| Identity | The fields that identify a record, so nara recognizes when two records describe the same thing. |
| Search fields | The fields nara embeds and indexes for search. Text search also matches a record’s key, and exact search its ID. |
| Usage | Knowledge (stored records), Parameter (Tool arguments), Return (Tool results). |
| Visibility | Visible types appear in the Knowledge app; System hidden types stay internal; Hidden (legacy) marks retired types. |
Versions
Section titled “Versions”Every published change creates a new version. References follow their Latest or Pinned mode, so you decide whether dependent types follow a change. When records already exist, nara migrates them to the new version and publishes it only when every record fits.