curl --request POST \
  --url https://api.lighton.ai/api/v3/content-types \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "action": "adopt",
  "content_types": [
    "legal",
    "finance"
  ]
}
'

{
  "content_types": [
    {
      "path": "legal",
      "code": "legal",
      "label": "Legal",
      "children": [
        {
          "path": "legal:contract",
          "code": "contract",
          "label": "Contract"
        }
      ]
    }
  ]
}

Facets

Define content types and attributes

Create, modify, and organise your company’s content-type trees and their attributes. Every action is idempotent, safe to re-run.

Two ways to build your schema:

Adopt from starter templates: import ready-made trees from the seed catalog (GET /api/v3/content-types/templates). Once adopted, they’re yours to customize.
Build from scratch: create your own trees, nodes, and attributes entirely with define_content_type and define_attribute.

You can mix both: adopt legal from the seeds, then add a custom compliance tree alongside it. Each tree is independent.

Actions:

adopt: import trees from the seed catalog (starter templates)
define_content_type: create or update a tree node (root or child)
undefine_content_type: delete a node and its entire subtree
define_attribute: add or update a metadata field on a node
undefine_attribute: remove a metadata field

Building a tree from scratch, example flow:

Create a root:

{"action": "define_content_type", "code": "compliance", "label": "Compliance"}

Add children:

{"action": "define_content_type", "parent_path": "compliance", "code": "audit-report", "label": "Audit Report"}

Add attributes:

{"action": "define_attribute", "content_type_path": "compliance", "name": "owner", "attribute_type": "text"}

Classify files via POST /api/v3/files/{id}/facets:

{"action": "classify", "content_type_path": "compliance:audit-report"}

Attributes are metadata fields you define on each node. Types: text, number, date, boolean, select, multi-select, rich-text. select and multi-select require choices. Attributes inherit down the tree: define jurisdiction on legal and it’s available on all children.

Description best practices: The description field on attributes feeds into the scope inference endpoint (POST /api/v3/content-types/scope), where it helps the LLM generate accurate search filters. Use keywords that signal the attribute’s nature:

Person names: include “name”, “examiner”, “inventor”, or “author” → enables *value* wildcard syntax hints
Codes/identifiers: include “code”, “classification”, or “identifier” → enables exact-match hints (don’t guess from topic keywords)

Multi-classification: A single document can be classified under multiple content types from different trees (e.g., both legal:contract:nda and compliance:audit-report).

Available seed roots for adopt: finance, healthcare, legal, manufacturing, tech

Requires admin-level access to modify the company’s schema.

POST

api

content-types

curl --request POST \
  --url https://api.lighton.ai/api/v3/content-types \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "action": "adopt",
  "content_types": [
    "legal",
    "finance"
  ]
}
'

{
  "content_types": [
    {
      "path": "legal",
      "code": "legal",
      "label": "Legal",
      "children": [
        {
          "path": "legal:contract",
          "code": "contract",
          "label": "Contract"
        }
      ]
    }
  ]
}

Authorizations

Authorization

string

header

required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

Request body for POST /api/v3/content-types.

Action-dispatched per FAC0012. Every action is idempotent.

Schema-side verb family:

adopt — bulk import from the Pydantic seed catalog.
define_content_type / undefine_content_type — CRUD on tree nodes.
define_attribute / undefine_attribute — CRUD on attribute columns.

Fields are validated per action in validate_fields_for_action — top-level optionality mirrors the union of action shapes, so consumers only need a single Pydantic class (friendly to drf-spectacular), but the validator enforces the narrow contract per action, the same pattern used by FileFacetActionRequest.

action

enum<string>

required

Available options:

adopt,

define_content_type,

undefine_content_type,

define_attribute,

undefine_attribute

content_types

string[]

parent_path

string | null

code

string | null

content_type_path

string | null

Colon-separated content type path.

label

string | null

Human-readable label for the node or attribute.

description

string | null

Optional descriptive text for the node or attribute.

inherit_attributes

boolean | null

Whether child content types inherit attributes from ancestors. Defaults to true on create.

name

string | null

Attribute identifier in snake_case.

attribute_type

string | null

Public attribute type. Supported values: text, number, date, boolean, select, multi-select, rich-text. Accepted aliases: multi_select, multiselect, rich_text, richtext.

required

boolean | null

Whether the attribute is required. Defaults to false.

choices

string[] | null

Required for select and multi-select attributes. Must be omitted for all other types.

Response

Action completed: existing node updated or already-present attribute replaced

ContentTypesListResponse
ContentTypeNodeResponse
AttributeDefResponse

content_types

ContentTypeNodeResponse · object[]

required

Show child attributes

can_edit

boolean | null

Overview

Intelligence

Document Processing

Administration

Define content types and attributes

Authorizations

Body

Response