UnrealMCP — AI Bridge for Unreal Engine 5

Enable or disable token debug estimation on MCP responses. When enabled, every command response shows a token debug header: [TOKEN DEBUG] command_name — ~1,289 tokens (5,156 chars) This appears at the top of every response so you can immediately see which commands consume the most tokens. Disabling clears accumulated stats. Args: enabled: True to enable, False to disable and clear stats.

Set material color on a mesh component. color: [R, G, B, A] values 0-1

Set a curve on a module input parameter. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter module_name: Name of the module input_name: Name of the input parameter script_usage: Stack the module is in curve_type: Curve type — "FloatCurve", "ColorCurve", "VectorCurve" keys: JSON array of curve keys, e.g. [{"time":0,"value":1},{"time":1,"value":0}] For color curves: [{"time":0,"r":1,"g":0,"b":0,"a":1}]

Set a dynamic input (random range, linked parameter, custom expression, or arbitrary script) on a module input. Uses FNiagaraStackGraphUtilities::SetDynamicInputForFunctionInput for the script path, which is the same code path the editor uses when picking a dynamic input from the stack UI. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter module_name: Name of the module input_name: Name of the input parameter script_usage: Stack the module is in dynamic_input_type: One of: - "random_range" / "uniform_random" — Ranged random (min_value, max_value) - "parameter_link" — Link to an existing parameter (parameter_name) - "custom_expression" — Custom HLSL expression node (expression) - "script" / "asset" — Arbitrary dynamic input script (dynamic_input_script_path) min_value / max_value: Values for random_range types (float or vector JSON object) parameter_name: Parameter name for parameter_link type expression: HLSL expression for custom_expression type dynamic_input_script_path: Asset path of a DynamicInput-usage UNiagaraScript (for dynamic_input_type="script"). Discover candidates via search_niagara_functions. suggested_name: Optional friendly name for the dynamic input in the stack UI pin_defaults: Optional {pin_name: default_value_string} dict applied to the new dynamic input node's own input pins after wiring

Set a property on an emitter within a Niagara System. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter property: Property name (e.g. "SimTarget", "FixedBounds", "LocalSpace") value: Property value as string

Enable or disable a module in an emitter's stack. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter module_name: Name of the module script_usage: Stack the module is in enabled: True to enable, false to disable

Set a static value on a module input parameter. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter module_name: Name of the module input_name: Name of the input parameter to set value: Value as string — scalars, vectors "(X,Y,Z)", colors "(R,G,B,A)", enums, bools script_usage: Stack the module is in

Set the timeline playback range in the Niagara editor. Controls how long the preview plays. If your effect dies at 0.02s, extend the range. For looping effects set a longer range (e.g. 5-10 seconds). Args: system_path: Path to the Niagara System asset range_end: End time in seconds (e.g. 5.0 for 5 second preview) range_start: Start time in seconds (default 0.0) frame_rate: Optional frame rate override (default 60)

Set a rapid iteration parameter value on a module. This is the primary way to change module input values like spawn rate, colors, forces, sizes, lifetimes, etc. Use get_niagara_rapid_iteration_parameters first to discover available parameters and their types. Value format depends on type: float: 5.0 int: 10 bool: true/false vector: {"x": 1, "y": 2, "z": 3} or [1, 2, 3] color: {"r": 1, "g": 0.5, "b": 0, "a": 1} or [1, 0.5, 0, 1] Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter module_name: Module name (e.g. "SpawnRate", "InitializeParticle", "GravityForce") input_name: Input parameter name (e.g. "SpawnRate", "Color", "Gravity") value: Value to set (type must match the parameter's type) script_usage: Stack the module is in — emitter_update, particle_spawn, particle_update, etc.

Set an attribute binding on an emitter's renderer. Binds a renderer input (e.g. color, size) to a particle/emitter attribute. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter binding_name: Binding slot name (e.g. "ColorBinding", "DynamicMaterialBinding", "PositionBinding", "SpriteRotationBinding") attribute: Attribute to bind to (e.g. "Particles.Color", "Particles.Scale") renderer_index: Index of the renderer (default 0)

Set a property on an emitter's renderer. Args: system_path: Path to the Niagara System asset emitter_name: Name of the target emitter property: Property name (e.g. "Material", "Alignment", "FacingMode", "SortOrder") value: Property value as string renderer_index: Index of the renderer (default 0)

Set HLSL source on a scratch pad module's Custom HLSL node, creating pins as needed. Pin creation uses exported schema APIs (UEdGraphSchema_Niagara::TypeDefinitionToPinType) plus reflection on the CustomHlsl UPROPERTY (SetCustomHlsl is not NIAGARAEDITOR_API exported). After changes the node's FNiagaraFunctionSignature is rebuilt from the pin list, mirroring UNiagaraNodeCustomHlsl::RebuildSignatureFromPins exactly. Args: system_path: Niagara System asset path module_name: Scratch pad module name containing the Custom HLSL node hlsl_code: Raw HLSL source. Reference pins by name (e.g. Result = Value * Scale;) inputs: Optional list of {"name": str, "type": str} — input pins to create outputs: Optional list of {"name": str, "type": str} — output pins to create clear_existing_pins: If true, remove all current non-Add pins before adding new ones

Batch-set details-panel properties on a Niagara script. Args: properties: Dict of property name -> value. Supported: Category (str), Description (str), Keywords (str), ModuleUsageBitmask (int), ProvidedDependencies (list[str]), RequiredDependencies (list[dict]), LibraryVisibility (str enum), bDeprecated (bool), DeprecationMessage (str), bExperimental (bool), ExperimentalMessage (str). system_path + module_name: Resolve scratch pad module script_path: OR resolve standalone script asset ModuleUsageBitmask combines ENiagaraScriptUsage values; e.g. particle spawn + update = (1<<3) | (1<<4).

Set a system-level property on a Niagara System via reflection. Supports any UPROPERTY on UNiagaraSystem: WarmupTime, WarmupTickDelta, bDeterminism, RandomSeed, bFixedBounds, FixedBounds, bFixedTickDelta, FixedTickDeltaTime, etc. Use exact C++ property name (PascalCase). Args: system_path: Path to the Niagara System asset property: Property name (e.g. "WarmupTime", "bDeterminism", "RandomSeed") value: Value as string (e.g. "2.0", "true", "42")

Set a user parameter value on a Niagara System asset or level actor instance. Provide either system_path (asset default) or actor_name (runtime override). Args: parameter_name: Name of the user parameter parameter_type: Type of the parameter (must match existing type) value: Value as string actor_name: Name of an actor in the level for runtime override system_path: Path to the system asset for default value

Set physics properties on a Blueprint component.

Set layout slot properties (anchors, offsets, padding, alignment, etc.).

Add or update a color in the StateTree theme. Args: asset_path: Content path to the StateTree asset name: Display name for the color color: Color as hex "#RRGGBB" or "R,G,B,A" floats (e.g. "0.5,0.2,0.8,1.0")

Set a property on a task, evaluator, or condition by GUID. Args: asset_path: Content path to the StateTree asset node_guid: GUID of the node property_name: Property to set property_value: New value (structs as dict, arrays as list, enums as string) target: "node" for the node struct itself, "instance" for instance data

Change the schema of a StateTree. Warning: existing nodes may become invalid. Args: asset_path: Content path to the StateTree asset schema_class: Schema class name (e.g. "MassStateTreeSchema")

Set a property on a state (Name, Type, SelectionBehavior, Enabled, Tag, Weight, etc.). Args: asset_path: Content path to the StateTree asset property_name: Property to set value: New value state: State index path state_name: State name state_guid: State GUID

Set a property on a transition (Trigger, Priority, TargetState, Delay, Enabled, etc.). Args: asset_path: Content path to the StateTree asset transition_index: Index of the transition in the state's Transitions array property_name: Property to set value: New value state: State index path state_name: State name state_guid: State GUID

Set the static mesh on a StaticMeshComponent in a Blueprint.

Set multiple widget properties. Same format as set_data_asset_properties.

Spawn an actor in the level. Args: name: Actor name actor_type: StaticMeshActor, PointLight, etc. location/rotation/scale: [x,y,z] arrays static_mesh: Path to static mesh asset (for mesh actors)

Spawn ANY AActor subclass by full path, short name, or Blueprint asset path. Replaces the hardcoded whitelist of spawn_actor. Resolves class_path in this order: 1. Full UClass path with '.' (e.g. "/Script/Engine.PostProcessVolume", "/Script/Engine.SkyAtmosphere") 2. Blueprint asset path (e.g. "/Game/Blueprints/BP_Miner") — loads the asset and uses its GeneratedClass. 3. Short name fallback — iterates loaded UClasses for a name match (handles "PostProcessVolume", "APostProcessVolume", "Sky_Sphere_C", etc.) Validates: must be AActor subclass, not abstract, not deprecated. Uses UEditorActorSubsystem::SpawnActorFromClass when available so Ctrl+Z restores. Args: class_path: Full path, BP asset path, or short class name. name: Optional outliner label + UObject name. location: [x, y, z] world location. rotation: [pitch, yaw, roll] degrees. scale: [x, y, z] scale.

Spawn a Blueprint actor instance in the level.

Spawn a Niagara System as an actor in the level. Args: system_path: Path to the Niagara System asset location: World location as "(X,Y,Z)" — defaults to origin rotation: Rotation as "(Pitch,Yaw,Roll)" — defaults to zero scale: Scale as "(X,Y,Z)" — defaults to (1,1,1) name: Optional actor label name auto_activate: If true, the system begins playing immediately

Navigate the Content Browser to show a specific asset.

Capture the editor to a PNG. mode "viewport" captures the level viewport; "window" captures the entire active editor window (material editor, BP graph, widget designer, etc.). Returns the file path, width, and height. Use Read tool to view the image.

Trace connections upstream and/or downstream from a specific node. Shows exactly what feeds into each input pin and what consumes each output pin, including connections to material outputs (BaseColor, Normal, etc.). Args: node_index: The node to trace from direction: "upstream" (what feeds in), "downstream" (what consumes output), or "both" (default) max_depth: How many hops to trace recursively (1 = immediate neighbors, 2+ = follow the chain). Max 50. Default 1.

Breadth-first trace of connections from a starting node through the graph. Answers "what feeds this node?" (upstream) and "where does this node's output go?" (downstream). Each visited node reports its depth, so the dependency chain is visible without dumping the whole graph. Args: system_path + module_name OR script_path: Graph resolver node_index / node_class / node_id: Starting node identifier direction: "upstream" | "downstream" | "both" (default) max_depth: Max BFS depth (default 8) pin_name: Optional starting-pin filter — only walk links on pins whose name contains this substring (e.g. "Vector Array")

Update specific fields on an existing row (partial update). Only fields present in data are modified; everything else keeps its value. data is a JSON string, e.g. '{"DisplayName": "Miner"}'.

Diagnose issues in a Material Function's internal graph. Reports: - unconnected_outputs: FunctionOutput nodes with nothing wired to their input - unused_inputs: FunctionInput nodes whose output nobody consumes - orphaned: internal nodes with no connections at all Returns healthy=true when no issues found.

Diagnose connection issues in a material graph. Categorises every node into: - orphaned: no inputs AND output not consumed — safe to delete - dead_ends: has inputs but output goes nowhere - missing_inputs: some input pins are empty - unconnected_inputs: ALL input pins are empty (node expects connections) - unconnected_outputs: output not consumed by anything Returns healthy=true when no orphaned or dead-end nodes exist.

Classify orphaned, dead-end, and missing-input nodes in a Niagara graph. Returns three arrays: orphaned — nodes with no incoming or outgoing links (excl. anchors) dead_ends — non-anchor nodes with inputs connected but no outputs consumed missing_inputs — nodes whose input pins have no link and no default value