Schema Types

Query (QueryRoot)

Field	Argument	Type	Description
hello		String!	Hello world demo
graph		Graph	Returns a graph
path		String!	Graph path relative to the root namespace (e.g. `"master"` or `"team/project/graph"`).
graphType		GraphType	Optional override for graph semantics — `EVENT` treats every update as a point-in-time event, `PERSISTENT` carries values forward until overwritten or deleted. Defaults to the stored graph's native type.
graphMetadata		MetaGraph	Returns lightweight metadata for a graph (node/edge counts, timestamps) without loading it. Requires at least INTROSPECT permission.
path		String!	Graph path relative to the root namespace.
updateGraph		MutableGraph!	Update graph query, has side effects to update graph state Returns:: GqlMutableGraph
path		String!	Graph path relative to the root namespace.
vectoriseGraph		Boolean!	Update graph query, has side effects to update graph state Returns:: GqlMutableGraph
path		String!	Graph path relative to the root namespace.
model		EmbeddingModel	Optional embedding model; defaults to OpenAI's standard model.
nodes		Template	Optional node-document template (which fields go into each node's text representation); defaults to the built-in template.
edges		Template	Optional edge-document template; defaults to the built-in template.
vectorisedGraph		VectorisedGraph	Create vectorised graph in the format used for queries Returns:: GqlVectorisedGraph
path		String!	Graph path relative to the root namespace.
namespaces		CollectionOfNamespace!	Returns all namespaces using recursive search Returns:: List of namespaces on root
namespace		Namespace!	Returns a specific namespace at a given path Returns:: Namespace or error if no namespace found
path		String!
root		Namespace!	Returns root namespace Returns:: Root namespace
plugins		QueryPlugin!	Returns a plugin.
receiveGraph		String!	Encodes graph and returns as string. Returns:: Base64 url safe encoded string
path		String!	Graph path relative to the root namespace.
version		String!	Version string of the running `raphtory-graphql` server build.

Mutation (MutRoot)

Field	Argument	Type	Description
plugins		MutationPlugin!	Returns a collection of mutation plugins.
deleteGraph		Boolean!	Delete graph from a path on the server.
path		String!	Graph path relative to the root namespace.
newGraph		Boolean!	Creates a new graph.
path		String!	Destination path relative to the root namespace.
graphType		GraphType!
moveGraph		Boolean!	Move graph from a path on the server to a new_path on the server.
path		String!	Current graph path relative to the root namespace.
newPath		String!	Destination path relative to the root namespace.
overwrite		Boolean	If true, allow replacing an existing graph at `newPath`; defaults to false.
copyGraph		Boolean!	Copy graph from a path on the server to a new_path on the server.
path		String!	Source graph path relative to the root namespace.
newPath		String!	Destination path relative to the root namespace.
overwrite		Boolean	If true, allow replacing an existing graph at `newPath`; defaults to false.
uploadGraph		String!	Upload a graph file from a path on the client using GQL multipart uploading. Returns:: name of the new graph
path		String!	Destination path relative to the root namespace.
graph		Upload!	Multipart upload of the serialised graph file.
overwrite		Boolean!	If true, replace any graph already at `path`.
sendGraph		String!	Send graph bincode as base64 encoded string. Returns:: path of the new graph
path		String!	Destination path relative to the root namespace.
graph		String!	Base64-encoded bincode of the serialised graph.
overwrite		Boolean!	If true, replace any graph already at `path`.
createNamespace		String!	Create an empty namespace at `path`. Creates any missing parent namespaces along the way. Requires WRITE permission on the parent namespace. Rejects paths that already host a graph or an existing namespace, and paths that fail validation. Returns:: the path of the created namespace
path		String!	Destination path relative to the root namespace.
deleteNamespace		Boolean!	Delete a namespace and all of its descendants (graphs and sub-namespaces). Requires WRITE permission on the parent namespace, on the namespace itself, and on every descendant graph and sub-namespace. Cached graphs at any deleted path are invalidated. Rejects empty and non-existent paths. Returns:: true on success
path		String!	Path to delete relative to the root namespace.
createSubgraph		String!	Returns a subgraph given a set of nodes from an existing graph in the server. Returns:: name of the new graph
parentPath		String!	Source graph path relative to the root namespace.
nodes		[NodeId!]!	Node ids to include in the subgraph.
newPath		String!	Destination path relative to the root namespace.
overwrite		Boolean!	If true, replace any graph already at `newPath`.
createIndex		Boolean!	(Experimental) Creates search index.
path		String!	Graph path relative to the root namespace.
indexSpec		IndexSpecInput	Optional spec selecting which node/edge property fields to index. Omit to index a default set.
inRam		Boolean!

Objects

CollectionOfMetaGraph

Collection of items

Field	Argument	Type	Description
list		[MetaGraph!]!	Returns a list of collection objects.
page		[MetaGraph!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
count		Int!	Returns a count of collection objects.

CollectionOfNamespace

Collection of items

Field	Argument	Type	Description
list		[Namespace!]!	Returns a list of collection objects.
page		[Namespace!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
count		Int!	Returns a count of collection objects.

CollectionOfNamespacedItem

Collection of items

Field	Argument	Type	Description
list		[NamespacedItem!]!	Returns a list of collection objects.
page		[NamespacedItem!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
count		Int!	Returns a count of collection objects.

Document

Document in a vector graph

Field	Argument	Type	Description
entity		DocumentEntity!	Entity associated with document.
content		String!	Content of the document.
embedding		[Float!]!	Similarity score with a specified query
score		Float!

Edge

Raphtory graph edge.

Field	Argument	Type	Description
defaultLayer		Edge!	Return a view of Edge containing only the default edge layer.
layers		Edge!	Returns a view of Edge containing all layers in the list of names. Errors if any of the layers do not exist.
names		[String!]!	Layer names to include.
excludeLayers		Edge!	Returns a view of Edge containing all layers except the excluded list of names. Errors if any of the layers do not exist.
names		[String!]!	Layer names to exclude.
layer		Edge!	Returns a view of Edge containing the specified layer. Errors if any of the layers do not exist.
name		String!	Layer name to include.
excludeLayer		Edge!	Returns a view of Edge containing all layers except the excluded layer specified. Errors if any of the layers do not exist.
name		String!	Layer name to exclude.
rolling		EdgeWindowSet!	Creates a WindowSet with the given window duration and optional step using a rolling window. A rolling window is a window that moves forward by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step (or window if no step is passed). e.g. "1 month and 1 day" will align at the start of the day. Note that passing a step larger than window while alignment_unit is not "Unaligned" may lead to some entries appearing before the start of the first window and/or after the end of the last window (i.e. not included in any window).
window		WindowDuration!	Width of each window. Pass either `{epoch: }` for a discrete number of milliseconds (e.g. `{epoch: 1000}` for 1 second), or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}` or `{duration: 2 hours and 30 minutes}`).
step		WindowDuration	Optional gap between the start of one window and the start of the next. Accepts the same `{epoch: }` or `{duration: }` values as `window`. Defaults to `window` — i.e. windows touch end-to-end with no overlap and no gap.
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step` (or `window` if no step is set).
expanding		EdgeWindowSet!	Creates a WindowSet with the given step size using an expanding window. An expanding window is a window that grows by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step. e.g. "1 month and 1 day" will align at the start of the day.
step		WindowDuration!	How much the window grows by on each step. Pass either `{epoch: }` for a discrete number of milliseconds, or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}`).
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step`.
window		Edge!	Creates a view of the Edge including all events between the specified start (inclusive) and end (exclusive). For persistent graphs, any edge which exists at any point during the window will be included. You may want to restrict this to only edges that are present at the end of the window using the is_valid function.
start		TimeInput!	Inclusive lower bound.
end		TimeInput!	Exclusive upper bound.
at		Edge!	Creates a view of the Edge including all events at a specified time.
time		TimeInput!	Instant to pin the view to.
latest		Edge!	View of this edge pinned to the graph's latest time — equivalent to `at(graph.latestTime)`. The edge's properties and metadata show their most recent values, and (for persistent graphs) validity is evaluated at that instant.
snapshotAt		Edge!	Creates a view of the Edge including all events that are valid at time. This is equivalent to before(time + 1) for Graph and at(time) for PersistentGraph.
time		TimeInput!	Instant at which entities must be valid.
snapshotLatest		Edge!	Creates a view of the Edge including all events that are valid at the latest time. This is equivalent to a no-op for Graph and latest() for PersistentGraph.
before		Edge!	Creates a view of the Edge including all events before a specified end (exclusive).
time		TimeInput!	Exclusive upper bound.
after		Edge!	Creates a view of the Edge including all events after a specified start (exclusive).
time		TimeInput!	Exclusive lower bound.
shrinkWindow		Edge!	Shrinks both the start and end of the window.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
shrinkStart		Edge!	Set the start of the window.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
shrinkEnd		Edge!	Set the end of the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
applyViews		Edge!	Takes a specified selection of views and applies them in given order.
views		[EdgeViewCollection!]!	Ordered list of view operations; each entry is a one-of variant (`window`, `layer`, `filter`, ...) applied to the running result.
earliestTime		EventTime!	Returns the earliest time of an edge.
firstUpdate		EventTime!	The timestamp of the first event in this edge's history (first update, first deletion, or anything in between). Differs from `earliestTime` in that `earliestTime` reports when the edge is first *valid*; `firstUpdate` reports when its history actually begins.
latestTime		EventTime!	Returns the latest time of an edge.
lastUpdate		EventTime!	The timestamp of the last event in this edge's history (last update, last deletion, or anything in between). Differs from `latestTime` in that `latestTime` reports when the edge is last *valid*; `lastUpdate` reports when its history actually ends.
time		EventTime!	Returns the time of an exploded edge. Errors on an unexploded edge.
start		EventTime!	Returns the start time for rolling and expanding windows for this edge. Returns none if no window is applied.
end		EventTime!	Returns the end time of the window. Returns none if no window is applied.
src		Node!	Returns the source node of the edge. Returns: Node:
dst		Node!	Returns the destination node of the edge. Returns: Node:
nbr		Node!	Returns the node at the other end of the edge (same as dst() for out-edges and src() for in-edges). Returns: Node:
id		[NodeId!]!	Returns the `[src, dst]` id pair of the edge. Each id is a `String` for string-indexed graphs or a non-negative `Int` for integer-indexed graphs.
properties		Properties!	Returns a view of the properties of the edge.
metadata		Metadata!	Returns the metadata of an edge.
layerNames		[String!]!	Returns the names of the layers that have this edge as a member.
layerName		String!	Returns the layer name of an exploded edge, errors on an edge.
explode		Edges!	Returns an edge object for each update within the original edge.
explodeLayers		Edges!	Returns an edge object for each layer within the original edge. Each new edge object contains only updates from the respective layers.
history		History!	Returns a History object with time entries for when an edge is added or change to an edge is made. Returns: History:
deletions		History!	Returns a history object with time entries for an edge's deletion times. Returns: History:
isValid		Boolean!	Checks if the edge is currently valid and exists at the current time. Returns: boolean
isActive		Boolean!	Checks if the edge is currently active and has at least one update within the current period. Returns: boolean
isDeleted		Boolean!	Checks if the edge is deleted at the current time. Returns: boolean
isSelfLoop		Boolean!	Returns true if the edge source and destination nodes are the same. Returns: boolean
filter		Edge!	Apply an edge filter in place, returning an edge view whose properties / metadata / history are restricted to the matching subset.
expr		EdgeFilter!	Composite edge filter (by property, layer, src/dst, etc.).

EdgeSchema

Describes edges between a specific pair of node types — the property and metadata keys seen on such edges, along with their observed value types. One EdgeSchema per (srcType, dstType) pair per layer.

Field	Argument	Type	Description
srcType		String!	Returns the type of source for these edges
dstType		String!	Returns the type of destination for these edges
properties		[PropertySchema!]!	Returns the list of property schemas for edges connecting these types of nodes
metadata		[PropertySchema!]!	Returns the list of metadata schemas for edges connecting these types of nodes

EdgeWindowSet

A lazy sequence of per-window views of a single edge, produced by edge.rolling / edge.expanding. Each entry is the edge as it exists in that window.

Field	Argument	Type	Description
count		Int!	Number of windows in this set. Materialising all windows is expensive for large graphs — prefer `page` over `list` when iterating.
page		[Edge!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Edge!]!	Materialise every window as a list. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.

Edges

A lazy collection of edges from a graph view. Supports the usual view transforms (window, layer, filter, ...), plus edge-specific ones like explode and explodeLayers, pagination, and sorting.

Field	Argument	Type	Description
defaultLayer		Edges!	Returns a collection containing only edges in the default edge layer.
layers		Edges!	Returns a collection containing only edges belonging to the listed layers.
names		[String!]!	Layer names to include.
excludeLayers		Edges!	Returns a collection containing edges belonging to all layers except the excluded list of layers.
names		[String!]!	Layer names to exclude.
layer		Edges!	Returns a collection containing edges belonging to the specified layer.
name		String!	Layer name to include.
excludeLayer		Edges!	Returns a collection containing edges belonging to all layers except the excluded layer specified.
name		String!	Layer name to exclude.
rolling		EdgesWindowSet!	Creates a WindowSet with the given window duration and optional step using a rolling window. A rolling window is a window that moves forward by step size at each iteration. Returns a collection of collections. This means that item in the window set is a collection of edges. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step (or window if no step is passed). e.g. "1 month and 1 day" will align at the start of the day. Note that passing a step larger than window while alignment_unit is not "Unaligned" may lead to some entries appearing before the start of the first window and/or after the end of the last window (i.e. not included in any window).
window		WindowDuration!	Width of each window. Pass either `{epoch: }` for a discrete number of milliseconds (e.g. `{epoch: 1000}` for 1 second), or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}` or `{duration: 2 hours and 30 minutes}`).
step		WindowDuration	Optional gap between the start of one window and the start of the next. Accepts the same `{epoch: }` or `{duration: }` values as `window`. Defaults to `window` — i.e. windows touch end-to-end with no overlap and no gap.
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step` (or `window` if no step is set).
expanding		EdgesWindowSet!	Creates a WindowSet with the given step size using an expanding window. An expanding window is a window that grows by step size at each iteration. Returns a collection of collections. This means that item in the window set is a collection of edges. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step. e.g. "1 month and 1 day" will align at the start of the day.
step		WindowDuration!	How much the window grows by on each step. Pass either `{epoch: }` for a discrete number of milliseconds, or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}`).
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step`.
window		Edges!	Creates a view of the Edge including all events between the specified start (inclusive) and end (exclusive).
start		TimeInput!	Inclusive lower bound.
end		TimeInput!	Exclusive upper bound.
at		Edges!	Creates a view of the Edge including all events at a specified time.
time		TimeInput!	Instant to pin the view to.
latest		Edges!	View showing only the latest state of each edge (equivalent to `at(latestTime)`).
snapshotAt		Edges!	Creates a view of the Edge including all events that are valid at time. This is equivalent to before(time + 1) for Graph and at(time) for PersistentGraph.
time		TimeInput!	Instant at which entities must be valid.
snapshotLatest		Edges!	Creates a view of the Edge including all events that are valid at the latest time. This is equivalent to a no-op for Graph and latest() for PersistentGraph.
before		Edges!	Creates a view of the Edge including all events before a specified end (exclusive).
time		TimeInput!	Exclusive upper bound.
after		Edges!	Creates a view of the Edge including all events after a specified start (exclusive).
time		TimeInput!	Exclusive lower bound.
shrinkWindow		Edges!	Shrinks both the start and end of the window.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
shrinkStart		Edges!	Set the start of the window.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
shrinkEnd		Edges!	Set the end of the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
applyViews		Edges!	Takes a specified selection of views and applies them in order given.
views		[EdgesViewCollection!]!	Ordered list of view operations; each entry is a one-of variant (`window`, `layer`, `filter`, ...) applied to the running result.
explode		Edges!	Expand each edge into one edge per update: if `A->B` has three updates, it becomes three `A->B` entries each at a distinct timestamp. Use this to iterate per-event rather than per-edge.
explodeLayers		Edges!	Returns an edge object for each layer within the original edge. Each new edge object contains only updates from the respective layers.
sorted		Edges!	Sort the edges. Multiple criteria are applied lexicographically (ties on the first key break to the second, etc.).
sortBys		[EdgeSortBy!]!	Ordered list of sort keys. Each entry chooses exactly one of `src` / `dst` / `time` / `property`, with an optional `reverse: true` to flip order.
start		EventTime!	Returns the start time of the window or none if there is no window.
end		EventTime!	Returns the end time of the window or none if there is no window.
count		Int!	Returns the number of edges. Returns: int:
page		[Edge!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Edge!]!	Returns a list of all objects in the current selection of the collection. You should filter the collection first then call list.
filter		Edges!	Narrow the collection to edges matching `expr`. The filter sticks to the returned view — every subsequent traversal through these edges (their properties, their endpoints' neighbours, etc.) continues to see the filtered scope. Useful when you want one scoping rule to apply across the whole query. E.g. restricting everything to a specific week: edges { filter(expr: {window: {start: 1234, end: 5678}}) { list { src { neighbours { list { name } } } } # neighbours still windowed } } Contrast with `select`, which applies here and is not carried through.
expr		EdgeFilter!	Composite edge filter (by property, layer, src/dst, etc.).
select		Edges!	Narrow the collection to edges matching `expr`, but only at this step — subsequent traversals out of these edges see the unfiltered graph again. Useful when you want different scopes at different hops. E.g. Monday's edges, then the neighbours of their endpoints on Tuesday, then *those* neighbours on Wednesday: edges { select(expr: {window: {...monday...}}) { list { src { select(expr: {window: {...tuesday...}}) { neighbours { select(expr: {window: {...wednesday...}}) { neighbours { list { name } } } } } } } } } Contrast with `filter`, which persists the scope through subsequent ops.
expr		EdgeFilter!	Composite edge filter (by property, layer, src/dst, etc.).

EdgesWindowSet

A lazy sequence of per-window edge collections, produced by edges.rolling / edges.expanding. Each entry is an Edges collection as it exists in that window.

Field	Argument	Type	Description
count		Int!	Number of windows in this set. Materialising all windows is expensive for large graphs — prefer `page` over `list` when iterating.
page		[Edges!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Edges!]!	Materialise every window as a list. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.

EventTime

Raphtory’s EventTime. Represents a unique timepoint in the graph’s history as (timestamp, event_id).

• timestamp: Number of milliseconds since the Unix epoch.
• event_id: ID used for ordering between equal timestamps.

Instances of EventTime may or may not contain time information. This is relevant for functions that may not return data (such as earliest_time and latest_time) because the data is unavailable. When empty, time operations (such as timestamp, datetime, and event_id) will return None.

Field	Argument	Type	Description
timestamp		Int	Get the timestamp in milliseconds since the Unix epoch.
eventId		Int	Get the event id for the EventTime. Used for ordering within the same timestamp.
datetime		String	Access a datetime representation of the EventTime as a String. Useful for converting millisecond timestamps into easily readable datetime strings. Optionally, a format string can be passed to format the output. Defaults to RFC 3339 if not provided (e.g., "2023-12-25T10:30:45.123Z"). Refer to chrono::format::strftime for formatting specifiers and escape sequences. Raises an error if a time conversion fails.
formatString		String	Optional format string for the rendered datetime. Uses `%`-style specifiers — for example `%Y-%m-%d` for `2024-01-15`, `%Y-%m-%d %H:%M:%S` for `2024-01-15 10:30:00`, or `%H:%M` for `10:30`. Defaults to RFC 3339 (e.g. `2024-01-15T10:30:45.123+00:00`) when omitted.

Graph

A view of a Raphtory graph. Every field here returns either data from the view or a derived view (window, layer, at, filter, ...) that you can keep chaining. Views are cheap — they don't copy the underlying data.

Field	Argument	Type	Description
uniqueLayers		[String!]!	Returns the names of all layers in the graphview. Distinct layer names observed in the current view — any layer that has at least one edge event visible here. Excludes layers that exist elsewhere in the graph but whose edges have been filtered out.
defaultLayer		Graph!	View restricted to the default layer — where nodes and edges end up when `addNode` / `addEdge` is called without a `layer` argument. Useful for separating "unlayered" base-graph events from named-layer ones.
layers		Graph!	View restricted to the named layers. Updates on any other layer are hidden; if that leaves a node or edge with no updates left, it disappears from the view.
names		[String!]!	Layer names to include.
excludeLayers		Graph!	View with the named layers hidden. Updates on those layers are removed; if that leaves a node or edge with no updates left, it disappears from the view.
names		[String!]!	Layer names to exclude.
layer		Graph!	View restricted to a single layer. Convenience form of `layers(names: [name])` — updates on any other layer are hidden, and entities with nothing left disappear.
name		String!	Layer name to include.
excludeLayer		Graph!	View with one layer hidden. Convenience form of `excludeLayers(names: [name])` — updates on that layer are removed, and entities with nothing left disappear.
name		String!	Layer name to exclude.
subgraph		Graph!	View restricted to a chosen set of nodes and the edges between them. Edges connecting a selected node to a non-selected node are hidden.
nodes		[NodeId!]!	Node ids to keep.
valid		Graph!	View containing only valid edges — for persistent graphs this drops edges whose most recent event is a deletion at the latest time of the current view (a later re-addition would keep them). On event graphs this is a no-op.
subgraphNodeTypes		Graph!	View restricted to nodes with the given node types.
nodeTypes		[String!]!	Node types to include.
excludeNodes		Graph!	View with a set of nodes removed (along with any edges touching them).
nodes		[NodeId!]!	Node ids to exclude.
rolling		GraphWindowSet!	Creates a rolling window with the specified window size and an optional step. A rolling window is a window that moves forward by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step (or window if no step is passed). e.g. "1 month and 1 day" will align at the start of the day. Note that passing a step larger than window while alignment_unit is not "Unaligned" may lead to some entries appearing before the start of the first window and/or after the end of the last window (i.e. not included in any window).
window		WindowDuration!	Width of each window. Pass either `{epoch: }` for a discrete number of milliseconds (e.g. `{epoch: 1000}` for 1 second), or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}` or `{duration: 2 hours and 30 minutes}`).
step		WindowDuration	Optional gap between the start of one window and the start of the next. Accepts the same `{epoch: }` or `{duration: }` values as `window`. Defaults to `window` — i.e. windows touch end-to-end with no overlap and no gap.
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step` (or `window` if no step is set).
expanding		GraphWindowSet!	Creates an expanding window with the specified step size. An expanding window is a window that grows by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step. e.g. "1 month and 1 day" will align at the start of the day.
step		WindowDuration!	How much the window grows by on each step. Pass either `{epoch: }` for a discrete number of milliseconds, or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}`).
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step`.
window		Graph!	Return a graph containing only the activity between start and end, by default raphtory stores times in milliseconds from the unix epoch.
start		TimeInput!	Inclusive lower bound.
end		TimeInput!	Exclusive upper bound.
at		Graph!	Creates a view including all events at a specified time.
time		TimeInput!	Instant to pin the view to.
latest		Graph!	Creates a view including all events at the latest time.
snapshotAt		Graph!	Create a view including all events that are valid at the specified time.
time		TimeInput!	Instant at which entities must be valid.
snapshotLatest		Graph!	Create a view including all events that are valid at the latest time.
before		Graph!	Create a view including all events before a specified end (exclusive).
time		TimeInput!	Exclusive upper bound.
after		Graph!	Create a view including all events after a specified start (exclusive).
time		TimeInput!	Exclusive lower bound.
shrinkWindow		Graph!	Shrink both the start and end of the window. The new bounds are taken as the intersection with the current window; this never widens the view.
start		TimeInput!	Proposed new start (TimeInput); ignored if before the current start.
end		TimeInput!	Proposed new end (TimeInput); ignored if after the current end.
shrinkStart		Graph!	Set the start of the window to the larger of the specified value or current start.
start		TimeInput!	Proposed new start (TimeInput); has no effect if it would widen the window.
shrinkEnd		Graph!	Set the end of the window to the smaller of the specified value or current end.
end		TimeInput!	Proposed new end (TimeInput); has no effect if it would widen the window.
created		Int!	Filesystem creation timestamp (epoch millis) of the graph's on-disk folder — i.e. when this graph was first saved to the server, not when its earliest event occurred. Use `earliestTime` for the latter.
lastOpened		Int!	Returns the graph's last opened timestamp according to system time.
lastUpdated		Int!	Returns the graph's last updated timestamp.
earliestTime		EventTime!	Returns the time entry of the earliest activity in the graph.
latestTime		EventTime!	Returns the time entry of the latest activity in the graph.
start		EventTime!	Returns the start time of the window. Errors if there is no window.
end		EventTime!	Returns the end time of the window. Errors if there is no window.
earliestEdgeTime		EventTime!	The earliest time at which any edge in this graph is valid. * `includeNegative` — if false, edge events with a timestamp `< 0` are skipped when computing the minimum. Defaults to true.
includeNegative		Boolean	If false, edge events with a timestamp `< 0` are skipped when computing the minimum. Defaults to true.
latestEdgeTime		EventTime!	The latest time at which any edge in this graph is valid.
includeNegative		Boolean	If false, edge events with a timestamp `< 0` are skipped when computing the maximum. Defaults to true.
countEdges		Int!	Returns the number of edges in the graph. Returns: int:
countTemporalEdges		Int!	Returns the number of temporal edges in the graph.
countNodes		Int!	Returns the number of nodes in the graph. Optionally takes a list of node ids to return a subset.
hasNode		Boolean!	Returns true if a node with the given id exists in this view.
name		NodeId!	Node id to look up.
hasEdge		Boolean!	Returns true if an edge exists between `src` and `dst` in this view, optionally restricted to a single layer.
src		NodeId!	Source node id.
dst		NodeId!	Destination node id.
layer		String	Optional; if provided, only checks whether the edge exists on this layer. If null or omitted, any layer counts.
node		Node	Look up a single node by id. Returns null if the node doesn't exist in this view.
name		NodeId!	Node id.
nodes		Nodes!	All nodes in this view, optionally narrowed by a filter.
select		NodeFilter	Optional node filter (by name, property, type, etc.). If omitted, every node in the view is returned.
edge		Edge	Look up a single edge by its endpoint ids. Returns null if no edge exists between `src` and `dst` in this view.
src		NodeId!	Source node id.
dst		NodeId!	Destination node id.
edges		Edges!	All edges in this view, optionally narrowed by a filter.
select		EdgeFilter	Optional edge filter (by property, layer, src/dst, etc.). If omitted, every edge in the view is returned.
properties		Properties!	Returns the properties of the graph.
metadata		Metadata!	Returns the metadata of the graph.
name		String!	Returns the graph name.
path		String!	Returns path of graph.
namespace		String!	Returns namespace of graph.
schema		GraphSchema!	Returns the graph schema.
algorithms		GraphAlgorithmPlugin!	Access registered graph algorithms (PageRank, shortest path, etc.) for this graph view. The set of available algorithms is defined by the plugin registry loaded at server startup.
sharedNeighbours		[Node!]!	Nodes that are neighbours of every node in `selectedNodes`. Returns the intersection of each selected node's neighbour set (undirected).
selectedNodes		[NodeId!]!	Node ids whose common neighbours you want. Returns an empty list if `selectedNodes` is empty or any id does not exist.
exportTo		Boolean!	Copy all nodes and edges of the current graph view into another already- existing graph stored on the server. The destination graph is preserved — this only adds; it does not replace.
path		String!	Destination graph path relative to the root namespace.
filter		Graph!	Returns a filtered view of the graph. Applies a mixed node/edge filter expression and narrows nodes, edges, and their properties to what matches.
expr		GraphFilter	Optional composite filter combining node, edge, property, and metadata conditions. If omitted, applies the identity filter (equivalent to no filtering).
filterNodes		Graph!	Returns a graph view restricted to nodes that match the given filter; edges are kept only if both endpoints survive.
expr		NodeFilter!	Composite node filter (by name, property, type, etc.).
filterEdges		Graph!	Returns a graph view restricted to edges that match the given filter. Nodes remain in the view even if all their edges are filtered out.
expr		EdgeFilter!	Composite edge filter (by property, layer, src/dst, etc.).
getIndexSpec		IndexSpec!	(Experimental) Get index specification.
searchNodes		[Node!]!	(Experimental) Searches for nodes which match the given filter expression. Uses Tantivy's exact search; requires the graph to have been indexed.
filter		NodeFilter!	Composite node filter (by name, property, type, etc.).
limit		Int!	Maximum number of nodes to return.
offset		Int!	Number of matches to skip before returning results.
searchEdges		[Edge!]!	(Experimental) Searches the index for edges which match the given filter expression. Uses Tantivy's exact search; requires the graph to have been indexed.
filter		EdgeFilter!	Composite edge filter (by property, layer, src/dst, etc.).
limit		Int!	Maximum number of edges to return.
offset		Int!	Number of matches to skip before returning results.
applyViews		Graph!	Apply a list of view operations in the given order and return the resulting graph view. Lets callers compose multiple view transforms (window, layer, filter, snapshot, ...) in a single call.
views		[GraphViewCollection!]!	Ordered list of view operations; each entry is a one-of variant applied to the running result.

GraphAlgorithmPlugin

Field	Argument	Type	Description
pagerank		[PagerankOutput!]!
iterCount		Int!
threads		Int
tol		Float
weight		String
shortest_path		[ShortestPathOutput!]!
source		String!
targets		[String!]!
direction		String

GraphSchema

Field	Argument	Type	Description
nodes		[NodeSchema!]!
layers		[LayerSchema!]!

GraphWindowSet

A lazy sequence of graph snapshots produced by rolling or expanding. Each entry is a Graph at a different window over time. Iterate via list / page (or count with count). Subsequent view ops apply per-window.

Field	Argument	Type	Description
count		Int!	Number of windows in this set. Materialising all windows is expensive for large graphs — prefer `page` over `list` when iterating.
page		[Graph!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Graph!]!	Materialise every window as a list. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.

History

History of updates for an object in Raphtory. Provides access to temporal properties.

Field	Argument	Type	Description
earliestTime		EventTime!	Get the earliest time entry associated with this history or None if the history is empty.
latestTime		EventTime!	Get the latest time entry associated with this history or None if the history is empty.
list		[EventTime!]!	List all time entries present in this history.
listRev		[EventTime!]!	List all time entries present in this history in reverse order.
page		[EventTime!]!	Fetch one page of EventTime entries with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
pageRev		[EventTime!]!	Fetch one page of EventTime entries with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page_rev(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
isEmpty		Boolean!	Returns True if the history is empty.
count		Int!	Get the number of entries contained in the history.
timestamps		HistoryTimestamp!	Returns a HistoryTimestamp object which accesses timestamps (milliseconds since the Unix epoch) instead of EventTime entries.
datetimes		HistoryDateTime!	Returns a HistoryDateTime object which accesses datetimes instead of EventTime entries. Useful for converting millisecond timestamps into easily readable datetime strings. Optionally, a format string can be passed to format the output. Defaults to RFC 3339 if not provided (e.g., "2023-12-25T10:30:45.123Z"). Refer to chrono::format::strftime for formatting specifiers and escape sequences.
formatString		String	Optional format string for the rendered datetime. Uses `%`-style specifiers — for example `%Y-%m-%d` for `2024-01-15`, `%Y-%m-%d %H:%M:%S` for `2024-01-15 10:30:00`, or `%H:%M` for `10:30`. Defaults to RFC 3339 (e.g. `2024-01-15T10:30:45.123+00:00`) when omitted.
eventId		HistoryEventId!	Returns a HistoryEventId object which accesses event ids of EventTime entries. They are used for ordering within the same timestamp.
intervals		Intervals!	Inter-event gap analysis for this history. The returned `Intervals` object exposes each gap (in milliseconds) between consecutive events, plus summary statistics — `min` / `max` / `mean` / `median` — and paginated access via `list` / `listRev` / `page` / `pageRev`.

HistoryDateTime

History object that provides access to datetimes instead of EventTime entries.

Field	Argument	Type	Description
list		[String!]!	List all datetimes formatted as strings. If filter_broken is set to True, time conversion errors will be ignored. If set to False, a TimeError will be raised on time conversion error. Defaults to False.
filterBroken		Boolean	If true, ignore unconvertible timestamps; if false, raise an error on the first conversion failure. Defaults to false.
listRev		[String!]!	List all datetimes formatted as strings in reverse chronological order. If filter_broken is set to True, time conversion errors will be ignored. If set to False, a TimeError will be raised on time conversion error. Defaults to False.
filterBroken		Boolean	If true, ignore unconvertible timestamps; if false, raise an error on the first conversion failure. Defaults to false.
page		[String!]!	Fetch one page of datetimes formatted as string with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). If filter_broken is set to True, time conversion errors will be ignored. If set to False, a TimeError will be raised on time conversion error. Defaults to False. For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
filterBroken		Boolean	If true, skip timestamps whose conversion fails; if false, raise an error on the first conversion failure. Defaults to false.
pageRev		[String!]!	Fetch one page of datetimes formatted as string in reverse chronological order with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). If filter_broken is set to True, time conversion errors will be ignored. If set to False, a TimeError will be raised on time conversion error. Defaults to False. For example, if page_rev(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
filterBroken		Boolean	If true, skip timestamps whose conversion fails; if false, raise an error on the first conversion failure. Defaults to false.

HistoryEventId

History object that provides access to event ids instead of EventTime entries.

Field	Argument	Type	Description
list		[Int!]!	List event ids.
listRev		[Int!]!	List event ids in reverse order.
page		[Int!]!	Fetch one page of event ids with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
pageRev		[Int!]!	Fetch one page of event ids in reverse chronological order with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page_rev(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).

HistoryTimestamp

History object that provides access to timestamps (milliseconds since the Unix epoch) instead of EventTime entries.

Field	Argument	Type	Description
list		[Int!]!	List all timestamps.
listRev		[Int!]!	List all timestamps in reverse order.
page		[Int!]!	Fetch one page of timestamps with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
pageRev		[Int!]!	Fetch one page of timestamps in reverse order with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page_rev(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).

IndexSpec

Field	Argument	Type	Description
nodeMetadata		[String!]!	Returns node metadata.
nodeProperties		[String!]!	Returns node properties.
edgeMetadata		[String!]!	Returns edge metadata.
edgeProperties		[String!]!	Returns edge properties.

Intervals

Provides access to the intervals between temporal entries of an object.

Field	Argument	Type	Description
list		[Int!]!	List time intervals between consecutive timestamps in milliseconds.
listRev		[Int!]!	List millisecond time intervals between consecutive timestamps in reverse order.
page		[Int!]!	Fetch one page of intervals between consecutive timestamps with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
pageRev		[Int!]!	Fetch one page of intervals between consecutive timestamps in reverse order with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
mean		Float	Compute the mean interval between consecutive timestamps. Returns None if fewer than 1 timestamp.
median		Int	Compute the median interval between consecutive timestamps. Returns None if fewer than 1 timestamp.
max		Int	Compute the maximum interval between consecutive timestamps. Returns None if fewer than 1 timestamp.
min		Int	Compute the minimum interval between consecutive timestamps. Returns None if fewer than 1 timestamp.

LayerSchema

Describes a single edge layer — its name and the per (srcType, dstType) edge schemas observed within it.

Field	Argument	Type	Description
name		String!	Returns the name of the layer with this schema
edges		[EdgeSchema!]!	Returns the list of edge schemas for this edge layer

MetaGraph

Lightweight summary of a stored graph — its name, path, counts, and filesystem timestamps — served without deserializing the full graph. Useful for listing what's available on the server before committing to a full load.

Field	Argument	Type	Description
name		String	Returns the graph name.
path		String!	Returns path of graph.
created		Int!	Returns the timestamp for the creation of the graph.
lastOpened		Int!	Returns the graph's last opened timestamp according to system time.
lastUpdated		Int!	Returns the graph's last updated timestamp.
nodeCount		Int!	Returns the number of nodes in the graph.
edgeCount		Int!	Returns the number of edges in the graph. Returns: int:
metadata		[Property!]!	Returns the metadata of the graph. Reads metadata without forcing a full graph load: from the in-memory cache if the graph is already loaded, otherwise directly from disk (parquet metadata for parquet-backed graphs, the `graph_props` segment for disk-backed graphs). This keeps `MetaGraph.metadata` cheap for namespace listings of many graphs.

Metadata

Constant key/value metadata attached to an entity (node, edge, or graph). Metadata has no timeline — each key maps to exactly one value for the lifetime of the entity. Separate from Properties, which carries time-varying data.

Field	Argument	Type	Description
get		Property	Look up a single metadata value by key. Returns null if no metadata with that key exists.
key		String!	The metadata name.
contains		Boolean!	Returns true if a metadata entry with the given key exists.
key		String!	The metadata name to look up.
keys		[String!]!	All metadata keys present on this entity.
values		[Property!]!	All metadata values as `{key, value}` entries.
keys		[String!]	Optional whitelist. If provided, only metadata with these keys is returned; if omitted, every metadata entry is returned.

MutableEdge

Write-side handle for a single edge — returned from addEdge or MutableGraph.edge. Supports adding updates, deletions, and attaching or updating metadata.

Field	Argument	Type	Description
success		Boolean!	Use to check if adding the edge was successful.
edge		Edge!	Get the non-mutable edge for querying.
src		MutableNode!	Get the mutable source node of the edge.
dst		MutableNode!	Get the mutable destination node of the edge.
delete		Boolean!	Mark this edge as deleted at the given time. Persistent graphs treat this as a tombstone (the edge becomes invalid from `time` onwards); event graphs simply log the deletion event.
time		TimeInput!	Time of the deletion.
layer		String	Optional layer name. If omitted, uses the layer the edge was originally added on (when called after `addEdge`).
addMetadata		Boolean!	Add metadata to this edge. Errors if any of the keys already exists — use `updateMetadata` to overwrite. If this is called after `addEdge`, the layer is inherited and does not need to be specified again.
properties		[PropertyInput!]!	List of `{key, value}` pairs to set as metadata.
layer		String	Optional layer name; defaults to the inherited layer.
updateMetadata		Boolean!	Update metadata of this edge, overwriting any existing values for the given keys. If this is called after `addEdge`, the layer is inherited and does not need to be specified again.
properties		[PropertyInput!]!	List of `{key, value}` pairs to upsert.
layer		String	Optional layer name; defaults to the inherited layer.
addUpdates		Boolean!	Append a property update to this edge at a specific time. If called after `addEdge`, the layer is inherited and does not need to be specified again.
time		TimeInput!	Time of the update.
properties		[PropertyInput!]	Optional `{key, value}` pairs attached to the event.
layer		String	Optional layer name; defaults to the inherited layer.

MutableGraph

Write-enabled handle for a graph. Obtained by calling updateGraph(path) on the root query with a path you have write permission for. Supports adding nodes and edges (individually or in batches), attaching properties/metadata, and looking up mutable node/edge handles. Use the read-only graph(path) resolver for queries.

Field	Argument	Type	Description
graph		Graph!	Read-only view of this graph — identical to what you'd get from `graph(path:)` on the query root. Use this when you want to compose queries on the graph you've just mutated. `graphType` lets you re-interpret the graph at query time (see `graph(path:)` for semantics); defaults to the stored graph's native type.
graphType		GraphType	Optional override for graph semantics — `EVENT` treats every update as a point-in-time event, `PERSISTENT` carries values forward until overwritten or deleted. Defaults to the stored graph's native type.
node		MutableNode	Look up an existing node for mutation. Returns null if the node doesn't exist; use `addNode` or `createNode` to create one.
name		NodeId!	Node id.
addNode		MutableNode!	Add a new node or append an update to an existing one. Upsert semantics: no error if the node already exists — properties and type are merged.
time		TimeInput!	Time of the event.
name		NodeId!	Node id.
properties		[PropertyInput!]	Optional property updates attached to this event.
nodeType		String	Optional node type to assign. If provided, sets the node's type at this event.
layer		String	Optional layer name. If omitted, the default layer is used.
createNode		MutableNode!	Create a new node or fail if it already exists. Strict alternative to `addNode` — use this when you want to detect collisions.
time		TimeInput!	Time of the create event.
name		NodeId!	Node id.
properties		[PropertyInput!]	Optional property updates attached to this event.
nodeType		String	Optional node type to assign. If provided, sets the node's type at this event.
layer		String	Optional layer name. If omitted, the default layer is used.
addNodes		Boolean!	Batch-add multiple nodes in one call. For each `NodeAddition`, applies every update it carries (time/properties pairs), then optionally sets its node type and adds any metadata. On partial failure, returns a `BatchFailures` error describing which entries failed and why; otherwise returns true.
nodes		[NodeAddition!]!	List of `NodeAddition` inputs, each specifying a node's name, optional type, layer, per-timestamp updates, and metadata.
edge		MutableEdge	Look up an existing edge for mutation. Returns null if no such edge exists.
src		NodeId!	Source node id.
dst		NodeId!	Destination node id.
addEdge		MutableEdge!	Add a new edge or append an update to an existing one. Upsert semantics: safe to call on an edge that already exists — creates missing endpoints if needed.
time		TimeInput!	Time of the event.
src		NodeId!	Source node id.
dst		NodeId!	Destination node id.
properties		[PropertyInput!]	Optional property updates attached to this event.
layer		String	Optional layer name. If omitted, the default layer is used.
addEdges		Boolean!	Batch-add multiple edges in one call. For each `EdgeAddition`, applies every update it carries, then adds any metadata. On partial failure, returns a `BatchFailures` error describing which entries failed; otherwise returns true.
edges		[EdgeAddition!]!	List of `EdgeAddition` inputs, each specifying an edge's `src`, `dst`, optional layer, per-timestamp updates, and metadata.
deleteEdge		MutableEdge!	Mark an edge as deleted at the given time. Persistent graphs treat this as a tombstone (the edge becomes invalid from `time` onwards); event graphs simply log the deletion event. Creates the edge first if it did not exist.
time		TimeInput!	Time of the deletion.
src		NodeId!	Source node id.
dst		NodeId!	Destination node id.
layer		String	Optional layer name. If omitted, the default layer is used.
addProperties		Boolean!	Add temporal properties to the graph itself (not a node or edge). Each call records a property update at `t`.
t		TimeInput!	Time of the update.
properties		[PropertyInput!]!	List of `{key, value}` pairs to set.
addMetadata		Boolean!	Add metadata to the graph itself. Errors if any of the keys already exists — use `updateMetadata` to overwrite.
properties		[PropertyInput!]!	List of `{key, value}` pairs to set as metadata.
updateMetadata		Boolean!	Update metadata of the graph itself, overwriting any existing values for the given keys.
properties		[PropertyInput!]!	List of `{key, value}` pairs to upsert.
flush		Boolean!	Persist any in-memory state for this graph to disk so other processes attaching a read-only handle observe up-to-date data.

MutableNode

Write-side handle for a single node — returned from addNode, createNode, or MutableGraph.node. Supports adding updates, setting node type, and attaching or updating metadata.

Field	Argument	Type	Description
success		Boolean!	Use to check if adding the node was successful.
node		Node!	Get the non-mutable Node.
addMetadata		Boolean!	Add metadata to this node. Errors if any of the keys already exists — use `updateMetadata` to overwrite.
properties		[PropertyInput!]!	List of `{key, value}` pairs to set as metadata.
setNodeType		Boolean!	Set this node's type. Errors if the node already has a non-default type and you're trying to change it.
newType		String!	Node-type name to assign.
updateMetadata		Boolean!	Update metadata of this node, overwriting any existing values for the given keys.
properties		[PropertyInput!]!	List of `{key, value}` pairs to upsert.
addUpdates		Boolean!	Append a property update to this node at a specific time.
time		TimeInput!	Time of the update.
properties		[PropertyInput!]	Optional `{key, value}` pairs attached to the event.
layer		String	Optional layer name. If omitted, the default layer is used.

MutationPlugin

Field	Argument	Type	Description
NoOps		String!

Namespace

A directory-like container for graphs and nested namespaces. Graphs are addressed by path (e.g. "team/project/graph"), and every segment except the last is a namespace. Use to browse what's stored on the server without loading any graph data.

Field	Argument	Type	Description
graphs		CollectionOfMetaGraph!	Graphs directly inside this namespace (excludes graphs in nested namespaces). Filtered by the caller's permissions — only graphs the caller is allowed to see are returned.
path		String!	Path of this namespace relative to the root namespace. Empty string for the root namespace itself.
parent		Namespace	Parent namespace, or null at the root.
children		CollectionOfNamespace!	Sub-namespaces directly inside this one (one level down, not recursive). Filtered by permissions.
items		CollectionOfNamespacedItem!	Everything in this namespace — sub-namespaces and graphs — as a single heterogeneous collection. Sub-namespaces are listed before graphs. Filtered by permissions.

Node

Raphtory graph node.

Field	Argument	Type	Description
id		NodeId!	Returns the unique id of the node — `String` for string-indexed graphs, non-negative `Int` for integer-indexed graphs.
name		String!	Returns the name of the node.
defaultLayer		Node!	Return a view of the node containing only the default layer.
layers		Node!	Return a view of node containing all layers specified.
names		[String!]!	Layer names to include.
excludeLayers		Node!	Returns a collection containing nodes belonging to all layers except the excluded list of layers.
names		[String!]!	Layer names to exclude.
layer		Node!	Returns a collection containing nodes belonging to the specified layer.
name		String!	Layer name to include.
excludeLayer		Node!	Returns a collection containing nodes belonging to all layers except the excluded layer.
name		String!	Layer name to exclude.
rolling		NodeWindowSet!	Creates a WindowSet with the specified window size and optional step using a rolling window. A rolling window is a window that moves forward by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step (or window if no step is passed). e.g. "1 month and 1 day" will align at the start of the day. Note that passing a step larger than window while alignment_unit is not "Unaligned" may lead to some entries appearing before the start of the first window and/or after the end of the last window (i.e. not included in any window).
window		WindowDuration!	Width of each window. Pass either `{epoch: }` for a discrete number of milliseconds (e.g. `{epoch: 1000}` for 1 second), or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}` or `{duration: 2 hours and 30 minutes}`).
step		WindowDuration	Optional gap between the start of one window and the start of the next. Accepts the same `{epoch: }` or `{duration: }` values as `window`. Defaults to `window` — i.e. windows touch end-to-end with no overlap and no gap.
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step` (or `window` if no step is set).
expanding		NodeWindowSet!	Creates a WindowSet with the specified step size using an expanding window. An expanding window is a window that grows by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step. e.g. "1 month and 1 day" will align at the start of the day.
step		WindowDuration!	How much the window grows by on each step. Pass either `{epoch: }` for a discrete number of milliseconds, or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}`).
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step`.
window		Node!	Create a view of the node including all events between the specified start (inclusive) and end (exclusive).
start		TimeInput!	Inclusive lower bound.
end		TimeInput!	Exclusive upper bound.
at		Node!	Create a view of the node including all events at a specified time.
time		TimeInput!	Instant to pin the view to.
latest		Node!	Create a view of the node including all events at the latest time.
snapshotAt		Node!	Create a view of the node including all events that are valid at the specified time.
time		TimeInput!	Instant at which entities must be valid.
snapshotLatest		Node!	Create a view of the node including all events that are valid at the latest time.
before		Node!	Create a view of the node including all events before specified end time (exclusive).
time		TimeInput!	Exclusive upper bound.
after		Node!	Create a view of the node including all events after the specified start time (exclusive).
time		TimeInput!	Exclusive lower bound.
shrinkWindow		Node!	Shrink a Window to a specified start and end time, if these are earlier and later than the current start and end respectively.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
shrinkStart		Node!	Set the start of the window to the larger of a specified start time and self.start().
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
shrinkEnd		Node!	Set the end of the window to the smaller of a specified end and self.end().
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
applyViews		Node!
views		[NodeViewCollection!]!
earliestTime		EventTime!	Returns the earliest time that the node exists.
firstUpdate		EventTime!	Returns the time of the first update made to the node.
latestTime		EventTime!	Returns the latest time that the node exists.
lastUpdate		EventTime!	Returns the time of the last update made to the node.
start		EventTime!	Gets the start time for the window. Errors if there is no window.
end		EventTime!	Gets the end time for the window. Errors if there is no window.
history		History!	Returns a history object for the node, with time entries for node additions and changes made to node.
edgeHistoryCount		Int!	Get the number of edge events for this node.
isActive		Boolean!	Check if the node is active and it's history is not empty.
nodeType		String	Returns the type of node.
properties		Properties!	Returns the properties of the node.
metadata		Metadata!	Returns the metadata of the node.
degree		Int!	Returns the number of unique counter parties for this node.
outDegree		Int!	Returns the number edges with this node as the source.
inDegree		Int!	Returns the number edges with this node as the destination.
inComponent		Nodes!
outComponent		Nodes!
edges		Edges!	Returns all connected edges.
select		EdgeFilter
outEdges		Edges!	Returns outgoing edges.
select		EdgeFilter
inEdges		Edges!	Returns incoming edges.
select		EdgeFilter
neighbours		PathFromNode!	Returns neighbouring nodes.
select		NodeFilter
inNeighbours		PathFromNode!	Returns the number of neighbours that have at least one in-going edge to this node.
select		NodeFilter
outNeighbours		PathFromNode!	Returns the number of neighbours that have at least one out-going edge from this node.
select		NodeFilter
filter		Node!
expr		NodeFilter!

NodeSchema

Describes nodes of a specific type in a graph — its property keys and observed value types (and, for string-valued properties, the set of distinct values seen). One NodeSchema per node type.

Field	Argument	Type	Description
typeName		String!	The node type this schema describes (e.g. `"person"`, `"org"`). Falls back to the default node type for untyped nodes.
properties		[PropertySchema!]!	Property schemas seen on nodes of this type — one entry per property key ever set on a node of this type, with its observed `PropertyType` and (for string-valued properties) the set of distinct values.
metadata		[PropertySchema!]!	Metadata schemas seen on nodes of this type — like `properties`, but covering metadata fields rather than temporal properties.

NodeWindowSet

A lazy sequence of per-window views of a single node, produced by node.rolling / node.expanding. Each entry is the node as it exists in that window.

Field	Argument	Type	Description
count		Int!	Number of windows in this set. Materialising all windows is expensive for large graphs — prefer `page` over `list` when iterating.
page		[Node!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Node!]!	Materialise every window as a list. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.

Nodes

A lazy collection of nodes from a graph view. Supports all the same view transforms as Graph (window, layer, filter, ...) plus pagination and sorting. Iterated via list / page / ids / count.

Field	Argument	Type	Description
defaultLayer		Nodes!	Return a view of the nodes containing only the default edge layer.
layers		Nodes!	Return a view of the nodes containing all layers specified.
names		[String!]!	Layer names to include.
excludeLayers		Nodes!	Return a view of the nodes containing all layers except those specified.
names		[String!]!	Layer names to exclude.
layer		Nodes!	Return a view of the nodes containing the specified layer.
name		String!	Layer name to include.
excludeLayer		Nodes!	Return a view of the nodes containing all layers except those specified.
name		String!	Layer name to exclude.
rolling		NodesWindowSet!	Creates a WindowSet with the specified window size and optional step using a rolling window. A rolling window is a window that moves forward by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step (or window if no step is passed). e.g. "1 month and 1 day" will align at the start of the day. Note that passing a step larger than window while alignment_unit is not "Unaligned" may lead to some entries appearing before the start of the first window and/or after the end of the last window (i.e. not included in any window).
window		WindowDuration!	Width of each window. Pass either `{epoch: }` for a discrete number of milliseconds (e.g. `{epoch: 1000}` for 1 second), or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}` or `{duration: 2 hours and 30 minutes}`).
step		WindowDuration	Optional gap between the start of one window and the start of the next. Accepts the same `{epoch: }` or `{duration: }` values as `window`. Defaults to `window` — i.e. windows touch end-to-end with no overlap and no gap.
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step` (or `window` if no step is set).
expanding		NodesWindowSet!	Creates a WindowSet with the specified step size using an expanding window. An expanding window is a window that grows by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step. e.g. "1 month and 1 day" will align at the start of the day.
step		WindowDuration!	How much the window grows by on each step. Pass either `{epoch: }` for a discrete number of milliseconds, or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}`).
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step`.
window		Nodes!	Create a view of the node including all events between the specified start (inclusive) and end (exclusive).
start		TimeInput!	Inclusive lower bound.
end		TimeInput!	Exclusive upper bound.
at		Nodes!	Create a view of the nodes including all events at a specified time.
time		TimeInput!	Instant to pin the view to.
latest		Nodes!	Create a view of the nodes including all events at the latest time.
snapshotAt		Nodes!	Create a view of the nodes including all events that are valid at the specified time.
time		TimeInput!	Instant at which entities must be valid.
snapshotLatest		Nodes!	Create a view of the nodes including all events that are valid at the latest time.
before		Nodes!	Create a view of the nodes including all events before specified end time (exclusive).
time		TimeInput!	Exclusive upper bound.
after		Nodes!	Create a view of the nodes including all events after the specified start time (exclusive).
time		TimeInput!	Exclusive lower bound.
shrinkWindow		Nodes!	Shrink both the start and end of the window.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
shrinkStart		Nodes!	Set the start of the window to the larger of a specified start time and self.start().
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
shrinkEnd		Nodes!	Set the end of the window to the smaller of a specified end and self.end().
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
typeFilter		Nodes!	Filter nodes by node type.
nodeTypes		[String!]!	Node-type names to keep.
applyViews		Nodes!	Apply a list of views in the given order and return the resulting nodes collection. Lets callers compose window, layer, filter, and snapshot operations in a single call.
views		[NodesViewCollection!]!	Ordered list of view operations; each entry is a one-of variant (`window`, `layer`, `filter`, etc.) applied to the running result.
sorted		Nodes!	Sort the nodes. Multiple criteria are applied lexicographically (ties on the first key break to the second, etc.).
sortBys		[NodeSortBy!]!	Ordered list of sort keys. Each entry chooses exactly one of `id` / `time` / `property`, with an optional `reverse: true` to flip order.
start		EventTime!	Returns the start time of the window. Errors if there is no window.
end		EventTime!	Returns the end time of the window. Errors if there is no window.
count		Int!	Number of nodes in the current view.
page		[Node!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Node!]!	Materialise every node in the view. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.
ids		[String!]!	Every node's id (name) as a flat list of strings. Rejected by the server when bulk list endpoints are disabled.
filter		Nodes!	Narrow the collection to nodes matching `expr`. The filter sticks to the returned view — every subsequent traversal through these nodes (their neighbours, edges, properties) continues to see the filtered scope. Useful when you want one scoping rule to apply across the whole query. E.g. restricting everything to a specific week: nodes { filter(expr: {window: {start: 1234, end: 5678}}) { list { neighbours { list { name } } } # neighbours still windowed } } Contrast with `select`, which applies here and is not carried through.
expr		NodeFilter!	Composite node filter (by name, property, type, etc.).
select		Nodes!	Narrow the collection to nodes matching `expr`, but only at this step — subsequent traversals out of these nodes see the unfiltered graph again. Useful when you want different scopes at different hops. E.g. nodes active on Monday, then their neighbours active on Tuesday, then *those* neighbours active on Wednesday: nodes { select(expr: {window: {...monday...}}) { list { neighbours { select(expr: {window: {...tuesday...}}) { list { neighbours { select(expr: {window: {...wednesday...}}) { list { name } } } } } } } } } Contrast with `filter`, which persists the scope through subsequent ops.
expr		NodeFilter!	Composite node filter (by name, property, type, etc.).

NodesWindowSet

A lazy sequence of per-window node collections, produced by nodes.rolling / nodes.expanding. Each entry is a Nodes collection as it exists in that window.

Field	Argument	Type	Description
count		Int!	Number of windows in this set. Materialising all windows is expensive for large graphs — prefer `page` over `list` when iterating.
page		[Nodes!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Nodes!]!	Materialise every window as a list. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.

PagerankOutput

PageRank score.

Field	Argument	Type	Description
name		String!
rank		Float!

PathFromNode

A collection of nodes anchored to a source node — the result of traversals like node.neighbours, inNeighbours, or outNeighbours. Supports all the usual view transforms (window, layer, filter, ...) and can be chained to walk further hops.

Field	Argument	Type	Description
layers		PathFromNode!	Returns a view of PathFromNode containing the specified layer, errors if the layer does not exist.
names		[String!]!	Layer names to include.
excludeLayers		PathFromNode!	Return a view of PathFromNode containing all layers except the specified excluded layers, errors if any of the layers do not exist.
names		[String!]!	Layer names to exclude.
layer		PathFromNode!	Return a view of PathFromNode containing the layer specified layer, errors if the layer does not exist.
name		String!	Layer name to include.
excludeLayer		PathFromNode!	Return a view of PathFromNode containing all layers except the specified excluded layers, errors if any of the layers do not exist.
name		String!	Layer name to exclude.
rolling		PathFromNodeWindowSet!	Creates a WindowSet with the given window size and optional step using a rolling window. A rolling window is a window that moves forward by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step (or window if no step is passed). e.g. "1 month and 1 day" will align at the start of the day. Note that passing a step larger than window while alignment_unit is not "Unaligned" may lead to some entries appearing before the start of the first window and/or after the end of the last window (i.e. not included in any window).
window		WindowDuration!	Width of each window. Pass either `{epoch: }` for a discrete number of milliseconds (e.g. `{epoch: 1000}` for 1 second), or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}` or `{duration: 2 hours and 30 minutes}`).
step		WindowDuration	Optional gap between the start of one window and the start of the next. Accepts the same `{epoch: }` or `{duration: }` values as `window`. Defaults to `window` — i.e. windows touch end-to-end with no overlap and no gap.
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step` (or `window` if no step is set).
expanding		PathFromNodeWindowSet!	Creates a WindowSet with the given step size using an expanding window. An expanding window is a window that grows by step size at each iteration. alignment_unit optionally aligns the windows to the specified unit. "Unaligned" can be passed for no alignment. If unspecified (i.e. by default), alignment is done on the smallest unit of time in the step. e.g. "1 month and 1 day" will align at the start of the day.
step		WindowDuration!	How much the window grows by on each step. Pass either `{epoch: }` for a discrete number of milliseconds, or `{duration: }` for a calendar duration (e.g. `{duration: 1 day}`).
alignmentUnit		AlignmentUnit	Optional anchor for window boundaries — pass `Unaligned` to disable, or one of the unit values (e.g. `Day`, `Hour`, `Minute`) to align edges to that calendar unit. Defaults to the smallest unit present in `step`.
window		PathFromNode!	Create a view of the PathFromNode including all events between a specified start (inclusive) and end (exclusive).
start		TimeInput!	Inclusive lower bound.
end		TimeInput!	Exclusive upper bound.
at		PathFromNode!	Create a view of the PathFromNode including all events at time.
time		TimeInput!	Instant to pin the view to.
snapshotLatest		PathFromNode!	Create a view of the PathFromNode including all events that are valid at the latest time.
snapshotAt		PathFromNode!	Create a view of the PathFromNode including all events that are valid at the specified time.
time		TimeInput!	Instant at which entities must be valid.
latest		PathFromNode!	Create a view of the PathFromNode including all events at the latest time.
before		PathFromNode!	Create a view of the PathFromNode including all events before the specified end (exclusive).
time		TimeInput!	Exclusive upper bound.
after		PathFromNode!	Create a view of the PathFromNode including all events after the specified start (exclusive).
time		TimeInput!	Exclusive lower bound.
shrinkWindow		PathFromNode!	Shrink both the start and end of the window.
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
shrinkStart		PathFromNode!	Set the start of the window to the larger of the specified start and self.start().
start		TimeInput!	Proposed new start (TimeInput); ignored if it would widen the window.
shrinkEnd		PathFromNode!	Set the end of the window to the smaller of the specified end and self.end().
end		TimeInput!	Proposed new end (TimeInput); ignored if it would widen the window.
typeFilter		PathFromNode!	Narrow this path to neighbours whose node type is in the given set.
nodeTypes		[String!]!	Node types to keep.
start		EventTime!	Returns the earliest time that this PathFromNode is valid or None if the PathFromNode is valid for all times.
end		EventTime!	Returns the latest time that this PathFromNode is valid or None if the PathFromNode is valid for all times.
count		Int!	Number of neighbour nodes reachable from the source in this view.
page		[Node!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[Node!]!	Materialise every neighbour node in the path. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.
ids		[String!]!	Every neighbour node's id (name) as a flat list of strings. Rejected by the server when bulk list endpoints are disabled.
applyViews		PathFromNode!	Takes a specified selection of views and applies them in given order.
views		[PathFromNodeViewCollection!]!	Ordered list of view operations; each entry is a one-of variant (`window`, `layer`, `filter`, ...) applied to the running result.
filter		PathFromNode!	Narrow the neighbour set to nodes matching `expr`. The filter sticks to the returned path — every subsequent traversal (further hops, edges, properties) continues to see the filtered scope. Useful when you want one scoping rule to apply across the whole query. E.g. restricting the whole traversal to a specific week: node(name: "A") { neighbours { filter(expr: {window: {...week...}}) { list { neighbours { list { name } } } # further hops still windowed } } } Contrast with `select`, which applies here and is not carried through.
expr		NodeFilter!	Composite node filter (by name, property, type, etc.).
select		PathFromNode!	Narrow the neighbour set to nodes matching `expr`, but only at this hop — further traversals out of these nodes see the unfiltered graph again. Useful when each hop needs a different scope. E.g. neighbours active on Monday, then *their* neighbours active on Tuesday: node(name: "A") { neighbours { select(expr: {window: {...monday...}}) { list { neighbours { select(expr: {window: {...tuesday...}}) { list { name } } } } } } } Contrast with `filter`, which persists the scope through subsequent ops.
expr		NodeFilter!	Composite node filter (by name, property, type, etc.).

PathFromNodeWindowSet

A lazy sequence of per-window neighbour sets, produced by neighbours.rolling / neighbours.expanding (or the in/out variants). Each entry is a PathFromNode scoped to that window.

Field	Argument	Type	Description
count		Int!	Number of windows in this set. Materialising all windows is expensive for large graphs — prefer `page` over `list` when iterating.
page		[PathFromNode!]!	Fetch one page with a number of items up to a specified limit, optionally offset by a specified amount. The page_index sets the number of pages to skip (defaults to 0). For example, if page(5, 2, 1) is called, a page with 5 items, offset by 11 items (2 pages of 5 + 1), will be returned.
limit		Int!	Maximum number of items to return on this page.
offset		Int	Extra items to skip on top of `pageIndex` paging (default 0).
pageIndex		Int	Zero-based page number; multiplies `limit` to determine where to start (default 0).
list		[PathFromNode!]!	Materialise every window as a list. Rejected by the server when bulk list endpoints are disabled; use `page` for paginated access instead.

Properties

All temporal properties of an entity (metadata is exposed separately). Look up individual properties via get / contains, enumerate via keys / values, or drop into temporal for time-aware accessors.

Field	Argument	Type	Description
get		Property	Look up a single property by key. Returns null if no property with that key exists in the current view.
key		String!	The property name.
contains		Boolean!	Returns true if a property with the given key exists in this view.
key		String!	The property name to look up.
keys		[String!]!	All property keys present in the current view. Does not include metadata — metadata is exposed separately via the entity's `metadata` field.
values		[Property!]!	Snapshot of property values, one `{key, value}` entry per property.
keys		[String!]	Optional whitelist. If provided, only properties with these keys are returned; if omitted or null, every property in the view is returned.
temporal		TemporalProperties!	The temporal-only view of these properties — excludes metadata (which has no history) and lets you drill into per-key timelines and aggregates.

Property

A single (key, value) property reading at a point in the graph view. The value is exposed both as a typed scalar (value) and as a human-readable string (asString).

Field	Argument	Type	Description
key		String!	The property key (name).
asString		String!	The property value rendered as a human-readable string (e.g. `"10"`, `"hello"`, `"2024-01-01T00:00:00Z"`). For programmatic access use `value`, which returns a typed scalar.
value		PropertyOutput!	The property value as a typed `PropertyOutput` scalar — numbers come back as numbers, booleans as booleans, strings as strings, etc.

PropertySchema

Field	Argument	Type	Description
key		String!
propertyType		String!
variants		[String!]!

PropertyTuple

A (time, value) pair — the output type of temporal-property accessors that need to report when a value was observed (e.g. min, max, median, orderedDedupe).

Field	Argument	Type	Description
time		EventTime!	The timestamp at which this value was recorded.
asString		String!	The value rendered as a human-readable string. For programmatic access use `value`, which returns a typed scalar.
value		PropertyOutput!	The value as a typed `PropertyOutput` scalar — numbers come back as numbers, booleans as booleans, etc.

QueryPlugin

Field	Argument	Type	Description
NoOps		String!

ShortestPathOutput

Field	Argument	Type	Description
target		String!
nodes		[String!]!

TemporalProperties

The temporal-only view of an entity's properties. Each entry is a TemporalProperty carrying the full timeline for that key — use this when you need per-update iteration, time-indexed lookups, or aggregates.

Field	Argument	Type	Description
get		TemporalProperty	Look up a single temporal property by key. Returns null if there's no temporal property with that key.
key		String!	The property name.
contains		Boolean!	Returns true if a temporal property with the given key exists.
key		String!	The property name to look up.
keys		[String!]!	All temporal-property keys present in this view.
values		[TemporalProperty!]!	All temporal properties, each as a `TemporalProperty` with its full timeline available. Use `history`, `values`, `latest`, `at`, etc. on each entry.
keys		[String!]	Optional whitelist. If provided, only temporal properties with these keys are returned; if omitted, every temporal property in the view is returned.

TemporalProperty

The full timeline of a single property key on one entity. Exposes every update (via values / history / orderedDedupe), point lookups (at, latest), and aggregates over the timeline (sum, mean, min, max, median, count).

Field	Argument	Type	Description
key		String!	The property key (name).
history		History!	Event history for this property — one entry per temporal update, in insertion order. Use this to navigate the full timeline: access the raw `timestamps` / `datetimes` / `eventId` lists, analyse gaps between updates via `intervals` (mean/median/min/max), ask `isEmpty`, or paginate the events.
values		[PropertyOutput!]!	All values this property has ever taken, in temporal order (one per update). Typed as `PropertyOutput` so numeric values stay numeric.
at		PropertyOutput	The value at or before time `t` (latest update on or before `t`). Returns null if no update exists on or before `t`.
t		TimeInput!	A TimeInput (epoch millis integer, RFC3339 string, or `{timestamp, eventId}` object).
latest		PropertyOutput	The most recent value, or null if the property has never been set in this view.
unique		[PropertyOutput!]!	The set of distinct values this property has ever taken (order not guaranteed).
orderedDedupe		[PropertyTuple!]!	Collapses runs of consecutive-equal updates into a single `(time, value)` pair.
latestTime		Boolean!	If true, each run is represented by its *last* timestamp; if false, by its *first*. Useful for compressing chatter in a timeline.
sum		PropertyOutput	Sum of all updates. Returns null if the dtype is not additive or the property is empty.
mean		PropertyOutput	Mean of all updates as an F64. Returns null if any value is non-numeric or the property is empty.
average		PropertyOutput	Alias for `mean` — same F64 average, same null cases.
min		PropertyTuple	Minimum `(time, value)` pair. Returns null if the dtype is not comparable or the property is empty.
max		PropertyTuple	Maximum `(time, value)` pair. Returns null if the dtype is not comparable or the property is empty.
median		PropertyTuple	Median `(time, value)` pair (lower median on even-length inputs). Returns null if the dtype is not comparable or the property is empty.
count		Int!	Number of updates recorded for this property in the current view.

VectorSelection

A working set of documents / nodes / edges built up via similarity searches on a VectorisedGraph. Selections are mutable: you can grow them with more hops (expand*), dereference the contents (nodes, edges, getDocuments), or start fresh with emptySelection.

Field	Argument	Type	Description
nodes		[Node!]!	Returns a list of nodes in the current selection.
edges		[Edge!]!	Returns a list of edges in the current selection.
getDocuments		[Document!]!	Returns a list of documents in the current selection.
addNodes		VectorSelection!	Add every document associated with the named nodes to the selection. Documents added this way receive a score of 0 (no similarity ranking).
nodes		[NodeId!]!	Node ids whose documents to include.
addEdges		VectorSelection!	Add every document associated with the named edges to the selection. Documents added this way receive a score of 0 (no similarity ranking).
edges		[InputEdge!]!	List of `{src, dst}` pairs identifying the edges.
expand		VectorSelection!	Grow the selection by including documents that are within `hops` of any document already in the selection. Two documents are 1 hop apart if they're on the same entity or on a connected node/edge pair.
hops		Int!	Number of expansion rounds (1 = direct neighbours).
window		VectorisedGraphWindow	Optional `{start, end}` to restrict expansion to entities active in that interval.
expandEntitiesBySimilarity		VectorSelection!	Iteratively expand the selection by similarity to a natural-language query. Each pass takes the one-hop neighbour set of the current selection and adds the highest-scoring entities (mixed nodes and edges); the loop continues until `limit` entities have been added.
query		String!	Natural-language search string; embedded by the server.
limit		Int!	Total number of entities to add across all passes.
window		VectorisedGraphWindow	Optional `{start, end}` to restrict matches to entities active in that interval.
expandNodesBySimilarity		VectorSelection!	Like `expandEntitiesBySimilarity` but restricted to nodes — iteratively add the highest-scoring adjacent nodes to the selection.
query		String!	Natural-language search string; embedded by the server.
limit		Int!	Total number of nodes to add across all passes.
window		VectorisedGraphWindow	Optional `{start, end}` to restrict matches to nodes active in that interval.
expandEdgesBySimilarity		VectorSelection!	Like `expandEntitiesBySimilarity` but restricted to edges — iteratively add the highest-scoring adjacent edges to the selection.
query		String!	Natural-language search string; embedded by the server.
limit		Int!	Total number of edges to add across all passes.
window		VectorisedGraphWindow	Optional `{start, end}` to restrict matches to edges active in that interval.

VectorisedGraph

A graph with embedded vector representations for its nodes and edges. Exposes similarity search over documents, nodes, and edges, plus selection building (emptySelection) and index maintenance (optimizeIndex).

Field	Argument	Type	Description
optimizeIndex		Boolean!	Rebuild (or incrementally update) the on-disk vector indexes for nodes and edges so subsequent similarity searches hit the fresh embeddings. Safe to call repeatedly; returns true on success.
emptySelection		VectorSelection!	Returns an empty selection of documents.
entitiesBySimilarity		VectorSelection!	Find the highest-scoring nodes *and* edges (mixed) by similarity to a natural-language query. The query is embedded server-side and matched against indexed entity vectors.
query		String!	Natural-language search string; embedded by the server.
limit		Int!	Maximum number of results to return.
window		VectorisedGraphWindow	Optional `{start, end}` to restrict matches to entities active in that interval.
nodesBySimilarity		VectorSelection!	Find the highest-scoring nodes by similarity to a natural-language query. The query is embedded server-side and matched against indexed node vectors.
query		String!	Natural-language search string; embedded by the server.
limit		Int!	Maximum number of nodes to return.
window		VectorisedGraphWindow	Optional `{start, end}` to restrict matches to nodes active in that interval.
edgesBySimilarity		VectorSelection!	Find the highest-scoring edges by similarity to a natural-language query. The query is embedded server-side and matched against indexed edge vectors.
query		String!	Natural-language search string; embedded by the server.
limit		Int!	Maximum number of edges to return.
window		VectorisedGraphWindow	Optional `{start, end}` to restrict matches to edges active in that interval.

Inputs

DegreeFilterNew

Field		Type	Description
direction		DegreeDirection!
where		PropCondition!

EdgeAddition

Field		Type	Description
src		NodeId!	Source node id (string or non-negative integer).
dst		NodeId!	Destination node id (string or non-negative integer).
layer		String	Layer.
metadata		[PropertyInput!]	Metadata.
updates		[TemporalPropertyInput!]

EdgeFilter

GraphQL input type for filtering edges.

EdgeFilter represents a composable boolean expression evaluated against edges in a graph. Filters can target:

• edge endpoints (source / destination nodes),
• edge properties and metadata,
• temporal scope (windows, snapshots, latest),
• layer membership,
• and structural edge state (active, valid, deleted, self-loop).

Filters can be combined recursively using logical operators (And, Or, Not).

Examples (GraphQL):

{
edges(filter: {
And: [
{ IsActive: true },
{ Property: { name: "weight", gt: 0.5 } }
]
}) {
src
dst
}
}

Field		Type	Description
src		NodeFilter	Applies a filter to the **source node** of the edge. The nested `NodeFilter` is evaluated against the source endpoint. Example: `{ Src: { Name: { contains: "alice" } } }`
dst		NodeFilter	Applies a filter to the **destination node** of the edge. The nested `NodeFilter` is evaluated against the destination endpoint. Example: `{ Dst: { Id: { eq: 42 } } }`
property		PropertyFilterNew	Filters an edge **property** by name and value. Applies to static or temporal properties depending on context. Example: `{ Property: { name: "weight", gt: 0.5 } }`
metadata		PropertyFilterNew	Filters an edge **metadata field**. Metadata is shared across all temporal versions of an edge. Example: `{ Metadata: { name: "source", eq: "imported" } }`
temporalProperty		PropertyFilterNew	Filters a **temporal edge property**. Used when the property value varies over time and must be evaluated within a temporal context. Example: `{ TemporalProperty: { name: "status", eq: "active" } }`
and		[EdgeFilter!]	Logical **AND** over multiple edge filters. All nested filters must evaluate to `true`. Example: `{ And: [ { IsActive: true }, { IsValid: true } ] }`
or		[EdgeFilter!]	Logical **OR** over multiple edge filters. At least one nested filter must evaluate to `true`. Example: `{ Or: [ { IsDeleted: true }, { IsSelfLoop: true } ] }`
not		EdgeFilter	Logical **NOT** over a nested edge filter. Negates the result of the wrapped filter. Example: `{ Not: { IsDeleted: true } }`
window		EdgeWindowExpr	Restricts edge evaluation to a **time window**. The window is inclusive of `start` and exclusive of `end`.
at		EdgeTimeExpr	Restricts edge evaluation to a **single point in time**.
before		EdgeTimeExpr	Restricts edge evaluation to times **strictly before** a given time.
after		EdgeTimeExpr	Restricts edge evaluation to times **strictly after** a given time.
latest		EdgeUnaryExpr	Evaluates edge predicates against the **latest available state**.
snapshotAt		EdgeTimeExpr	Evaluates edge predicates against a **snapshot** of the graph at a specific time.
snapshotLatest		EdgeUnaryExpr	Evaluates edge predicates against the **most recent snapshot** of the graph.
layers		EdgeLayersExpr	Restricts evaluation to edges belonging to one or more **layers**. Example: `{ Layers: { values: ["fire_nation", "air_nomads"] } }`
isActive		Boolean	Matches edges that have at least one event in the current view/window. When `true`, only active edges are matched.
isValid		Boolean	Matches edges that are structurally valid (i.e. not deleted) in the current view/window.
isDeleted		Boolean	Matches edges that have been deleted in the current view/window.
isSelfLoop		Boolean	Matches edges that are **self-loops** (source node == destination node).

EdgeLayersExpr

Restricts edge evaluation to one or more layers and applies a nested EdgeFilter.

Used by GqlEdgeFilter::Layers.

Field		Type	Description
names		[String!]!	Layer names to include.
expr		EdgeFilter!	Filter evaluated within the layer-restricted view.

EdgeSortBy

Field		Type	Description
reverse		Boolean	Reverse order
src		Boolean	Source node
dst		Boolean	Destination
time		SortByTime	Time
property		String	Property

EdgeTimeExpr

Restricts edge evaluation to a single time bound and applies a nested EdgeFilter.

Used by At, Before, and After edge filters.

Field		Type	Description
time		TimeInput!	Reference time for the operation.
expr		EdgeFilter!	Filter evaluated within the restricted time scope.

EdgeUnaryExpr

Applies a unary edge-view operation and then evaluates a nested EdgeFilter.

Used by Latest and SnapshotLatest edge filters.

Field		Type	Description
expr		EdgeFilter!	Filter evaluated after applying the unary operation.

EdgeViewCollection

Field		Type	Description
defaultLayer		Boolean	Contains only the default layer.
latest		Boolean	Latest time.
snapshotLatest		Boolean	Snapshot at latest time.
snapshotAt		TimeInput	Snapshot at specified time.
layers		[String!]	List of included layers.
excludeLayers		[String!]	List of excluded layers.
excludeLayer		String	Single excluded layer.
window		Window	Window between a start and end time.
at		TimeInput	View at a specified time.
before		TimeInput	View before a specified time (end exclusive).
after		TimeInput	View after a specified time (start exclusive).
shrinkWindow		Window	Shrink a Window to a specified start and end time.
shrinkStart		TimeInput	Set the window start to a specified time.
shrinkEnd		TimeInput	Set the window end to a specified time.
edgeFilter		EdgeFilter	Edge filter

EdgeWindowExpr

Restricts edge evaluation to a time window and applies a nested EdgeFilter.

Used by GqlEdgeFilter::Window.

The window is inclusive of start and exclusive of end.

Field		Type	Description
start		TimeInput!	Window start time (inclusive).
end		TimeInput!	Window end time (exclusive).
expr		EdgeFilter!	Filter evaluated within the restricted window.

EdgesViewCollection

Field		Type	Description
defaultLayer		Boolean	Contains only the default layer.
latest		Boolean	Latest time.
snapshotLatest		Boolean	Snapshot at latest time.
snapshotAt		TimeInput	Snapshot at specified time.
layers		[String!]	List of included layers.
excludeLayers		[String!]	List of excluded layers.
excludeLayer		String	Single excluded layer.
window		Window	Window between a start and end time.
at		TimeInput	View at a specified time.
before		TimeInput	View before a specified time (end exclusive).
after		TimeInput	View after a specified time (start exclusive).
shrinkWindow		Window	Shrink a Window to a specified start and end time.
shrinkStart		TimeInput	Set the window start to a specified time.
shrinkEnd		TimeInput	Set the window end to a specified time.
edgeFilter		EdgeFilter	Edge filter

EmbeddingModel

Field		Type	Description
openAI		OpenAIConfig	OpenAI embedding models or compatible providers

GraphFilter

GraphQL input type for restricting a graph view.

GraphFilter controls the evaluation scope for subsequent node/edge filters: - time windows (Window) - time points (At) - open-ended ranges (Before, After) - latest evaluation (Latest) - snapshots (SnapshotAt, SnapshotLatest) - layer membership (Layers)

These filters can be nested via the expr field on the corresponding *Expr input objects to form pipelines.

Field		Type	Description
window		GraphWindowExpr	Restrict evaluation to a time window (inclusive start, exclusive end).
at		GraphTimeExpr	Restrict evaluation to a single point in time.
before		GraphTimeExpr	Restrict evaluation to times strictly before the given time.
after		GraphTimeExpr	Restrict evaluation to times strictly after the given time.
latest		GraphUnaryExpr	Evaluate against the latest available state.
snapshotAt		GraphTimeExpr	Evaluate against a snapshot of the graph at a given time.
snapshotLatest		GraphUnaryExpr	Evaluate against the most recent snapshot of the graph.
layers		GraphLayersExpr	Restrict evaluation to one or more layers.

GraphLayersExpr

Graph view restriction by layer membership, optionally chaining another GraphFilter.

Used by GqlGraphFilter::Layers.

Field		Type	Description
names		[String!]!	Layer names to include.
expr		GraphFilter	Optional nested filter applied after the layer restriction.

GraphTimeExpr

Graph view restriction to a single time bound, optionally chaining another GraphFilter.

Used by At, Before, and After graph filters.

Example: { At: { time: 5, expr: { Layers: { names: ["L1"] } } } }

Field		Type	Description
time		TimeInput!	Reference time for the operation.
expr		GraphFilter	Optional nested filter applied after the time restriction.

GraphUnaryExpr

Graph view restriction that takes only a nested expression.

Used for unary view operations like Latest and SnapshotLatest.

Field		Type	Description
expr		GraphFilter	Optional nested filter applied after the unary operation.

GraphViewCollection

Field		Type	Description
defaultLayer		Boolean	Contains only the default layer.
layers		[String!]	List of included layers.
excludeLayers		[String!]	List of excluded layers.
excludeLayer		String	Single excluded layer.
subgraph		[NodeId!]	Subgraph nodes.
subgraphNodeTypes		[String!]	Subgraph node types.
excludeNodes		[NodeId!]	List of excluded nodes.
valid		Boolean	Valid state.
window		Window	Window between a start and end time.
at		TimeInput	View at a specified time.
latest		Boolean	View at the latest time.
snapshotAt		TimeInput	Snapshot at specified time.
snapshotLatest		Boolean	Snapshot at latest time.
before		TimeInput	View before a specified time (end exclusive).
after		TimeInput	View after a specified time (start exclusive).
shrinkWindow		Window	Shrink a Window to a specified start and end time.
shrinkStart		TimeInput	Set the window start to a specified time.
shrinkEnd		TimeInput	Set the window end to a specified time.
nodeFilter		NodeFilter	Node filter.
edgeFilter		EdgeFilter	Edge filter.

GraphWindowExpr

Graph view restriction to a time window, optionally chaining another GraphFilter.

Used by GqlGraphFilter::Window.

• start and end define the window (inclusive start, exclusive end).
• expr optionally nests another graph filter to apply within this window.

Example (GraphQL):

{ Window: { start: 0, end: 10, expr: { Layers: { names: ["A"] } } } }

Field		Type	Description
start		TimeInput!	Window start time (inclusive).
end		TimeInput!	Window end time (exclusive).
expr		GraphFilter	Optional nested filter applied after the window restriction.

IndexSpecInput

Field		Type	Description
nodeProps		PropsInput!	Node properties.
edgeProps		PropsInput!	Edge properties.

InputEdge

Field		Type	Description
src		NodeId!	Source node id (string or non-negative integer).
dst		NodeId!	Destination node id (string or non-negative integer).

NodeAddition

Field		Type	Description
name		NodeId!	Node id (string or non-negative integer).
nodeType		String	Node type.
metadata		[PropertyInput!]	Metadata.
updates		[TemporalPropertyInput!]	Updates.
layer		String	Layer.

NodeFieldCondition

Boolean expression over a built-in node field (ID, name, or type).

This is used by NodeFieldFilterNew.where_ when filtering a specific NodeField.

Supports comparisons, string predicates, and set membership. (Presence checks and aggregations are handled via property filters instead.)

Field		Type	Description
eq		Value	Equality.
ne		Value	Inequality.
gt		Value	Greater-than.
ge		Value	Greater-than-or-equal.
lt		Value	Less-than.
le		Value	Less-than-or-equal.
startsWith		Value	String prefix match.
endsWith		Value	String suffix match.
contains		Value	Substring match.
notContains		Value	Negated substring match.
isIn		Value	Set membership.
isNotIn		Value	Negated set membership.

NodeFieldFilterNew

Filters a built-in node field (id, name, type) using a NodeFieldCondition.

Example (GraphQL):

{ Node: { field: NodeName, where: { Contains: "ali" } } }

Field		Type	Description
field		NodeField!	Which built-in field to filter.
where		NodeFieldCondition!	Condition applied to the selected field. Exposed as `where` in GraphQL.

NodeFilter

GraphQL input type for filtering nodes.

NodeFilter represents a composable boolean expression evaluated against nodes in a graph. Filters can target:

• built-in node fields (Node / NodeFieldFilterNew),
• node properties and metadata,
• temporal properties,
• temporal scope (windows, snapshots, latest),
• and layer membership,
• plus node state predicates (e.g. IsActive).

Filters can be combined recursively using logical operators (And, Or, Not).

Field		Type	Description
node		NodeFieldFilterNew	Filters a built-in node field (ID, name, or type).
property		PropertyFilterNew	Filters a node property by name and condition.
degree		DegreeFilterNew	Filters a node's degree (in, out, or total) by a condition.
metadata		PropertyFilterNew	Filters a node metadata field by name and condition. Metadata is shared across all temporal versions of a node.
temporalProperty		PropertyFilterNew	Filters a temporal node property by name and condition. Used when the property value varies over time and must be evaluated within a temporal context.
and		[NodeFilter!]	Logical AND over multiple node filters.
or		[NodeFilter!]	Logical OR over multiple node filters.
not		NodeFilter	Logical NOT over a nested node filter.
window		NodeWindowExpr	Restricts evaluation to a time window (inclusive start, exclusive end).
at		NodeTimeExpr	Restricts evaluation to a single point in time.
before		NodeTimeExpr	Restricts evaluation to times strictly before the given time.
after		NodeTimeExpr	Restricts evaluation to times strictly after the given time.
latest		NodeUnaryExpr	Evaluates predicates against the latest available node state.
snapshotAt		NodeTimeExpr	Evaluates predicates against a snapshot of the graph at a given time.
snapshotLatest		NodeUnaryExpr	Evaluates predicates against the most recent snapshot of the graph.
layers		NodeLayersExpr	Restricts evaluation to nodes belonging to one or more layers.
isActive		Boolean	Matches nodes that have at least one event in the current view/window. When `true`, only active nodes are matched.

NodeLayersExpr

Restricts node evaluation to one or more layers and applies a nested NodeFilter.

Used by GqlNodeFilter::Layers.

Field		Type	Description
names		[String!]!	Layer names to include.
expr		NodeFilter!	Filter evaluated within the layer-restricted view.

NodeSortBy

Field		Type	Description
reverse		Boolean	Reverse order
id		Boolean	Unique Id
time		SortByTime	Time
property		String	Property

NodeTimeExpr

Restricts node evaluation to a single time bound and applies a nested NodeFilter.

Used by At, Before, and After node filters.

Field		Type	Description
time		TimeInput!	Reference time for the operation.
expr		NodeFilter!	Filter evaluated within the restricted time scope.

NodeUnaryExpr

Applies a unary node-view operation and then evaluates a nested NodeFilter.

Used by Latest and SnapshotLatest node filters.

Field		Type	Description
expr		NodeFilter!	Filter evaluated after applying the unary operation.

NodeViewCollection

Field		Type	Description
defaultLayer		Boolean	Contains only the default layer.
latest		Boolean	View at the latest time.
snapshotLatest		Boolean	Snapshot at latest time.
snapshotAt		TimeInput	Snapshot at specified time.
layers		[String!]	List of included layers.
excludeLayers		[String!]	List of excluded layers.
excludeLayer		String	Single excluded layer.
window		Window	Window between a start and end time.
at		TimeInput	View at a specified time.
before		TimeInput	View before a specified time (end exclusive).
after		TimeInput	View after a specified time (start exclusive).
shrinkWindow		Window	Shrink a Window to a specified start and end time.
shrinkStart		TimeInput	Set the window start to a specified time.
shrinkEnd		TimeInput	Set the window end to a specified time.
nodeFilter		NodeFilter	Node filter.

NodeWindowExpr

Restricts node evaluation to a time window and applies a nested NodeFilter.

Used by GqlNodeFilter::Window.

The window is inclusive of start and exclusive of end.

Field		Type	Description
start		TimeInput!	Window start time (inclusive).
end		TimeInput!	Window end time (exclusive).
expr		NodeFilter!	Filter evaluated within the restricted window.

NodesViewCollection

Field		Type	Description
defaultLayer		Boolean	Contains only the default layer.
latest		Boolean	View at the latest time.
snapshotLatest		Boolean	Snapshot at latest time.
layers		[String!]	List of included layers.
excludeLayers		[String!]	List of excluded layers.
excludeLayer		String	Single excluded layer.
window		Window	Window between a start and end time.
at		TimeInput	View at a specified time.
snapshotAt		TimeInput	Snapshot at specified time.
before		TimeInput	View before a specified time (end exclusive).
after		TimeInput	View after a specified time (start exclusive).
shrinkWindow		Window	Shrink a Window to a specified start and end time.
shrinkStart		TimeInput	Set the window start to a specified time.
shrinkEnd		TimeInput	Set the window end to a specified time.
nodeFilter		NodeFilter	Node filter.
typeFilter		[String!]	List of types.

ObjectEntry

Field		Type	Description
key		String!	Key.
value		Value!	Value.

OpenAIConfig

Field		Type	Description
model		String!
apiBase		String
apiKeyEnv		String
orgId		String
projectId		String

PathFromNodeViewCollection

Field		Type	Description
latest		Boolean	Latest time.
snapshotLatest		Boolean	Latest snapshot.
snapshotAt		TimeInput	Time.
layers		[String!]	List of layers.
excludeLayers		[String!]	List of excluded layers.
excludeLayer		String	Single layer to exclude.
window		Window	Window between a start and end time.
at		TimeInput	View at a specified time.
before		TimeInput	View before a specified time (end exclusive).
after		TimeInput	View after a specified time (start exclusive).
shrinkWindow		Window	Shrink a Window to a specified start and end time.
shrinkStart		TimeInput	Set the window start to a specified time.
shrinkEnd		TimeInput	Set the window end to a specified time.

PropCondition

Boolean expression over a property value.

PropCondition is used inside PropertyFilterNew.where to describe how a property’s value should be matched.

It supports: - comparisons (Eq, Gt, Le, …), - string predicates (Contains, StartsWith, …), - set membership (IsIn, IsNotIn), - presence checks (IsSome, IsNone), - boolean composition (And, Or, Not), - and list/aggregate qualifiers (First, Sum, Len, …).

Notes: - Value is interpreted according to the property’s type. - Aggregators/qualifiers like Sum and Len apply when the underlying property is list-like or aggregatable (depending on your engine rules).

Field		Type	Description
eq		Value	Equality: property value equals the given value.
ne		Value	Inequality: property value does not equal the given value.
gt		Value	Greater-than: property value is greater than the given value.
ge		Value	Greater-than-or-equal: property value is >= the given value.
lt		Value	Less-than: property value is less than the given value.
le		Value	Less-than-or-equal: property value is <= the given value.
startsWith		Value	String prefix match against the property's string representation.
endsWith		Value	String suffix match against the property's string representation.
contains		Value	Substring match against the property's string representation.
notContains		Value	Negated substring match against the property's string representation.
isIn		Value	Set membership: property value is contained in the given list of values.
isNotIn		Value	Negated set membership: property value is not contained in the given list of values.
isSome		Boolean	Presence check: property value is present (not null/missing). When set to `true`, requires the property to exist.
isNone		Boolean	Absence check: property value is missing / null. When set to `true`, requires the property to be missing.
and		[PropCondition!]	Logical AND over nested conditions.
or		[PropCondition!]	Logical OR over nested conditions.
not		PropCondition	Logical NOT over a nested condition.
first		PropCondition	Applies the nested condition to the **first** element of a list-like property.
last		PropCondition	Applies the nested condition to the **last** element of a list-like property.
any		PropCondition	Requires that **any** element of a list-like property matches the nested condition.
all		PropCondition	Requires that **all** elements of a list-like property match the nested condition.
sum		PropCondition	Applies the nested condition to the **sum** of a numeric list-like property.
avg		PropCondition	Applies the nested condition to the **average** of a numeric list-like property.
min		PropCondition	Applies the nested condition to the **minimum** element of a list-like property.
max		PropCondition	Applies the nested condition to the **maximum** element of a list-like property.
len		PropCondition	Applies the nested condition to the **length** of a list-like property.

PropertyFilterNew

Filters an entity property or metadata field by name and condition.

This input is used by both node and edge filters when targeting a specific property key (or metadata key) and applying a PropCondition.

Fields: - name: The property key to query. - where_: The condition to apply to that property’s value.

Example (GraphQL):

{ Property: { name: "weight", where: { Gt: 0.5 } } }

Field		Type	Description
name		String!	Property (or metadata) key.
where		PropCondition!	Condition applied to the property value. Exposed as `where` in GraphQL.

PropertyInput

Field		Type	Description
key		String!	Key.
value		Value!	Value.

PropsInput

Field		Type	Description
all		AllPropertySpec	All properties and metadata.
some		SomePropertySpec	Some properties and metadata.

SomePropertySpec

SomePropertySpec object containing lists of metadata and property names.

Field		Type	Description
metadata		[String!]!	List of metadata.
properties		[String!]!	List of properties.

Template

Field		Type	Description
enabled		Boolean	The default template.
custom		String	A custom template.

TemporalPropertyInput

Field		Type	Description
time		TimeInput!	Time of the update — accepts the same forms as `TimeInput` (epoch millis Int, RFC3339 string, or `{timestamp, eventId}` object).
properties		[PropertyInput!]	Properties.

Value

Field		Type	Description
u8		Int	8 bit unsigned integer.
u16		Int	16 bit unsigned integer.
u32		Int	32 bit unsigned integer.
u64		Int	64 bit unsigned integer.
i32		Int	32 bit signed integer.
i64		Int	64 bit signed integer.
f32		Float	32 bit float.
f64		Float	64 bit float.
str		String	String.
bool		Boolean	Boolean.
list		[Value!]	List.
object		[ObjectEntry!]	Object.
dtime		String	Timezone-aware datetime.
ndtime		String	Naive datetime (no timezone).
decimal		String	BigDecimal number (string representation, e.g. "3.14159" or "123e-5").

VectorisedGraphWindow

Field		Type	Description
start		TimeInput!	Inclusive lower bound of the search window.
end		TimeInput!	Exclusive upper bound of the search window.

Window

Field		Type	Description
start		TimeInput!	Window start time.
end		TimeInput!	Window end time.

WindowDuration

Field		Type	Description
duration		String	Duration of window period. Choose from:
epoch		Int	Time.

Enums

AlignmentUnit

Alignment unit used to align window boundaries.

Value	Description
UNALIGNED
MILLISECOND
SECOND
MINUTE
HOUR
DAY
WEEK
MONTH
YEAR

AllPropertySpec

Value	Description
ALL	All properties and metadata.
ALL_METADATA	All metadata.
ALL_PROPERTIES	All properties.

DegreeDirection

Filters nodes by computed degree with a directional scope.

DegreeFilterNew lets callers filter on: - inbound degree (IN), - outbound degree (OUT), - or total degree (BOTH).

The selected degree is compared using the where condition.

Example (GraphQL):

{ Degree: { direction: BOTH, where: { Gt: 10 } } }

Value	Description
IN
OUT
BOTH

GraphType

Value	Description
PERSISTENT	Persistent.
EVENT	Event.

NodeField

Value	Description
NODE_ID	Node ID field. Represents the graph’s node identifier (numeric or string-backed in the API).
NODE_NAME	Node name field. Represents the human-readable node name (string).
NODE_TYPE	Node type field. Represents the optional node type assigned at node creation (string).

SortByTime

Value	Description
LATEST	Latest time
EARLIEST	Earliest time

Scalars

Boolean

The Boolean scalar type represents true or false.

Float

The Float scalar type represents signed double-precision fractional values as specified by IEEE 754.

Int

The Int scalar type represents non-fractional signed whole numeric values. Int can represent values between -(2^31) and 2^31 - 1.

NodeId

Identifier for a node — either a string ("alice") or a non-negative integer (42). Use whichever form matches how the graph was indexed when nodes were added.

PropertyOutput

String

The String scalar type represents textual data, represented as UTF-8 character sequences. The String type is most often used by GraphQL to represent free-form human-readable text.

TimeInput

Input for primary time component. Expects Int, DateTime formatted String, or Object { timestamp, eventId } where the timestamp is either an Int or a DateTime formatted String, and eventId is a non-negative Int. Valid string formats are RFC3339, RFC2822, %Y-%m-%d, %Y-%m-%dT%H:%M:%S%.3f, %Y-%m-%dT%H:%M:%S%, %Y-%m-%d %H:%M:%S%.3f and %Y-%m-%d %H:%M:%S%.

Internally wraps InputTime so write paths (addNode, addEdge, addProperties, etc.) can preserve auto-increment of event_id when only a timestamp is given. Pass the object form {timestamp, eventId} to lock the event_id explicitly.

Upload

A multipart file upload

Unions

DocumentEntity

Entity associated with document.

Type	Description
Node	Raphtory graph node.
Edge	Raphtory graph edge.

NamespacedItem

Type	Description
Namespace	A directory-like container for graphs and nested namespaces. Graphs are addressed by path (e.g. `"team/project/graph"`), and every segment except the last is a namespace. Use to browse what's stored on the server without loading any graph data.
MetaGraph	Lightweight summary of a stored graph — its name, path, counts, and filesystem timestamps — served without deserializing the full graph. Useful for listing what's available on the server before committing to a full load.