Update API

README.md

@@ -136,7 +136,8 @@ csv_data = "name,email,age\nJohn,john@example.com,30\nJane,jane@example.com,25"
 db.records.import_csv(
     label='USER',
     data=csv_data,
-    options={'returnResult': True, 'suggestTypes': True},
+    # skipEmptyValues: treat empty cells ("" / []) as unset instead of storing them (0/False are kept)
+    options={'returnResult': True, 'suggestTypes': True, 'skipEmptyValues': True},
     parse_config={'header': True, 'skipEmptyLines': True, 'dynamicTyping': True},
 )
 ```

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "rushdb"
-version = "2.3.0"
+version = "2.6.0"
 description = "RushDB Python SDK — memory layer for AI agents and modern apps"
 authors = [
     {name = "RushDB Team", email = "hi@rushdb.com"}

src/rushdb/api/ai.py

@@ -1,6 +1,6 @@
 """AI API for RushDB Python SDK.
 
-Provides methods for graph ontology exploration, semantic vector search,
+Provides methods for graph schema exploration, semantic vector search,
 and embedding index management.
 """
 
@@ -123,7 +123,7 @@ class AIAPI(BaseAPI):
     Example::
 
         >>> db = RushDB(api_key="...")
-        >>> ontology = db.ai.get_ontology()
+        >>> schema = db.ai.get_schema()
         >>> results = db.ai.search({"query": "fast cars", "propertyName": "description"})
         >>> db.ai.indexes.create({"propertyName": "description", "label": "Book"})
     """
@@ -132,12 +132,12 @@ def __init__(self, client: "RushDB"):
         super().__init__(client)
         self.indexes = _AIIndexesNamespace(client)
 
-    def get_ontology(
+    def get_schema(
         self,
         params: Optional[Dict[str, Any]] = None,
         transaction: Optional[Union[Transaction, str]] = None,
     ) -> ApiResponse:
-        """Return the full graph ontology as structured JSON.
+        """Return the full graph schema as structured JSON.
 
         Each item contains the label name, record count, properties with value
         ranges/samples, and cross-label relationships with direction.
@@ -149,13 +149,13 @@ def get_ontology(
 
         Args:
             params: Optional filter. Pass ``{"labels": ["Label1"]}`` to scope
-                the ontology to specific labels only. Pass ``{"force": True}``
-                to bypass the 1-hour ontology cache and trigger a full
+                the schema to specific labels only. Pass ``{"force": True}``
+                to bypass the 1-hour schema cache and trigger a full
                 recalculation.
             transaction: Optional transaction context.
 
         Returns:
-            ApiResponse: Response whose ``data`` is a list of ontology items.
+            ApiResponse: Response whose ``data`` is a list of schema items.
         """
         headers: Dict[str, str] = {}
         if transaction is not None:
@@ -165,20 +165,20 @@ def get_ontology(
             headers["x-transaction-id"] = tx_id
 
         response = self.client._make_request(
-            "POST", "/ai/ontology", params or {}, headers=headers or None
+            "POST", "/ai/schema", params or {}, headers=headers or None
         )
         return ApiResponse(
             data=response.get("data"),
             success=response.get("success", True),
             total=response.get("total"),
         )
 
-    def get_ontology_markdown(
+    def get_schema_markdown(
         self,
         params: Optional[Dict[str, Any]] = None,
         transaction: Optional[Union[Transaction, str]] = None,
     ) -> ApiResponse:
-        """Return the full graph ontology as compact Markdown tables.
+        """Return the full graph schema as compact Markdown tables.
 
         Token-efficient representation intended for direct LLM consumption.
         Includes labels with counts, properties with types and value
@@ -191,7 +191,7 @@ def get_ontology_markdown(
         Args:
             params: Optional filter. Pass ``{"labels": ["Label1"]}`` to scope
                 the output to specific labels only. Pass ``{"force": True}``
-                to bypass the 1-hour ontology cache and trigger a full
+                to bypass the 1-hour schema cache and trigger a full
                 recalculation.
             transaction: Optional transaction context.
 
@@ -206,7 +206,7 @@ def get_ontology_markdown(
             headers["x-transaction-id"] = tx_id
 
         response = self.client._make_request(
-            "POST", "/ai/ontology/md", params or {}, headers=headers or None
+            "POST", "/ai/schema/md", params or {}, headers=headers or None
         )
         return ApiResponse(
             data=response.get("data"),

src/rushdb/api/records.py

@@ -172,6 +172,8 @@ def create(
                 Available options:
                 - returnResult (bool): Whether to return the created record data. Defaults to True.
                 - suggestTypes (bool): Whether to automatically suggest data types. Defaults to True.
+                - skipEmptyValues (bool): Treat empty strings ("") and empty arrays ([]) as unset —
+                  no property is created. 0 and False are kept. Defaults to False.
             vectors (Optional[List[Dict[str, Any]]]): Optional pre-computed embedding vectors
                 to write alongside the record. Each entry must contain at least
                 ``propertyName`` and ``vector``; ``similarityFunction`` is required when
@@ -231,7 +233,9 @@ def create_many(
                 record. Nested objects/arrays are not supported here — raises
                 ``ValueError`` if any item contains a nested object or list.
             options: Optional write options forwarded as-is to the server
-                (e.g. ``suggestTypes``, ``mergeBy``, ``mergeStrategy``, etc.).
+                (e.g. ``suggestTypes``, ``skipEmptyValues``, ``mergeBy``,
+                ``mergeStrategy``, etc.). ``skipEmptyValues`` treats empty strings
+                and empty arrays as unset (``0``/``False`` are kept); defaults to False.
             vectors: Optional per-row inline vectors for external embedding indexes.
                 ``vectors[i]`` is applied to ``data[i]``. Each element is a list of
                 vector entry dicts: ``[{"propertyName": str, "vector": List[float],

src/rushdb/api/relationship_patterns.py

@@ -10,7 +10,7 @@
 
 
 class RelationshipPatternsAPI(BaseAPI):
-    """Review and manage relationship patterns inferred from project ontology.
+    """Review and manage relationship patterns inferred from project schema.
 
     Accessed via ``db.relationships.patterns``.
     """
@@ -27,12 +27,12 @@ def _wrap(response: Dict[str, Any]) -> ApiResponse:
         )
 
     def list(self) -> ApiResponse:
-        """List inferred patterns, ontology relationships, and analysis status."""
+        """List inferred patterns, schema relationships, and analysis status."""
         response = self.client._make_request("GET", "/relationships/patterns")
         return self._wrap(response)
 
     def analyze(self) -> ApiResponse:
-        """Queue ontology analysis to generate relationship pattern suggestions."""
+        """Queue schema analysis to generate relationship pattern suggestions."""
         response = self.client._make_request(
             "POST", "/relationships/patterns/analyze", {}
         )

src/rushdb/models/api_response.py

@@ -19,7 +19,7 @@ class ApiResponse(Generic[T]):
         total: Optional total count (e.g. for list endpoints).
 
     Example:
-        >>> response = db.ai.get_ontology_markdown()
+        >>> response = db.ai.get_schema_markdown()
         >>> schema_md = response.data          # str
         >>> response = db.ai.indexes.find()
         >>> indexes = response.data            # list[dict]

src/rushdb/models/property.py

@@ -21,8 +21,9 @@ class DatetimeObject(TypedDict, total=False):
 NumberValue = Union[float, int]
 StringValue = str
 
-# Property types
-PropertyType = Literal["boolean", "datetime", "null", "number", "string"]
+# Property types. A None/null value means the field is unset; `None` is still accepted as an
+# input value in `PropertyWithValue.value` below.
+PropertyType = Literal["boolean", "datetime", "number", "string"]
 
 
 class Property(TypedDict):