Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/davidbuenov/dbv-mcp-server/llms.txt

Use this file to discover all available pages before exploring further.

The Unreal Engine 5.8 MCP server exposes three built-in management tools at the top level. These are always available immediately after initialization — no toolset discovery is needed to call them. They are the entry point for exploring and invoking every other capability registered in the engine.

​
list_toolsets

Returns a plain-text listing of all toolsets currently registered in the Unreal Engine Toolset Registry, including each toolset’s name and a short description. Arguments: none Returns: text — a newline-delimited list of toolset entries, each formatted as: 
- <FullyQualifiedToolsetName>: <Description>
The proxy parses this output (regex - ([A-Za-z0-9\._]+):) to obtain the toolset names it needs to call describe_toolset during cache reconstruction. Example call 
{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "list_toolsets",
    "arguments": {}
  }
}
Example response (SSE body) 
event: message
data: {"jsonrpc":"2.0","id":2,"result":{"content":[{"type":"text","text":"- editor_toolset.toolsets.scene.SceneTools: Scene management tools\n- editor_toolset.toolsets.actor.ActorTools: Actor tools\n- ToolsetRegistry.AgentSkillToolset: Agent skill management\n..."}]}}

​
describe_toolset

Returns detailed schemas for every tool inside a specific toolset — including tool names, descriptions, and full JSON Schema inputSchema objects. Use this to understand what arguments a tool expects before calling it.
​
toolset_name
string
required
The fully-qualified toolset name as returned by list_toolsets. Example: "ToolsetRegistry.AgentSkillToolset" or "editor_toolset.toolsets.scene.SceneTools".
Returns: text containing a JSON object: 
{
  "tools": [
    {
      "name": "SpawnActor",
      "description": "Spawns an actor in the level.",
      "inputSchema": {
        "type": "object",
        "properties": { ... },
        "required": [...]
      }
    }
  ]
}
Example call 
{
  "jsonrpc": "2.0",
  "id": 3,
  "method": "tools/call",
  "params": {
    "name": "describe_toolset",
    "arguments": {
      "toolset_name": "ToolsetRegistry.AgentSkillToolset"
    }
  }
}

​
call_tool

Executes a tool by name inside a specified toolset. This is the primary dispatch mechanism used by the proxy when translating namespaced tool calls (e.g. editor_toolset.toolsets.scene.SceneTools.SpawnActor) into the native call format.
​
tool_name
string
required
The unqualified name of the tool to call — without any toolset prefix. Example: "SpawnActor", "ListSkills".
​
toolset_name
string
The fully-qualified name of the toolset that contains the tool. Omit to call a top-level MCP tool by name. Use list_toolsets to discover valid values.
​
arguments
object
Arguments to pass to the tool. Must match the tool’s inputSchema as returned by describe_toolset. Defaults to {}.
Example call 
{
  "jsonrpc": "2.0",
  "id": 4,
  "method": "tools/call",
  "params": {
    "name": "call_tool",
    "arguments": {
      "toolset_name": "editor_toolset.toolsets.scene.SceneTools",
      "tool_name": "SpawnActor",
      "arguments": {
        "actor_type": { "refPath": "/Script/Engine.PointLight" },
        "xform": {
          "location": { "x": 0, "y": 0, "z": 300 },
          "rotation": { "pitch": 0, "yaw": 0, "roll": 0 },
          "scale":    { "x": 1, "y": 1, "z": 1 }
        }
      }
    }
  }
}

​
AgentSkillToolset

ToolsetRegistry.AgentSkillToolset is the one toolset that Unreal Engine 5.8 registers automatically at startup (via ToolsetRegistrySubsystem::Initialize()). It manages AgentSkill assets — structured knowledge units that persist AI instructions inside a UE project. Call these tools via call_tool with toolset_name: "ToolsetRegistry.AgentSkillToolset".

​
ListSkills

Returns a summary of all AgentSkill assets in the project. Arguments: none ({}) Returns: object mapping SkillPath → Description 
{
  "jsonrpc": "2.0",
  "id": 5,
  "method": "tools/call",
  "params": {
    "name": "call_tool",
    "arguments": {
      "toolset_name": "ToolsetRegistry.AgentSkillToolset",
      "tool_name": "ListSkills",
      "arguments": {}
    }
  }
}

​
GetSkills

Returns detailed instructions for one or more AgentSkill assets by path.
​
skillPaths
string[]
required
Array of full asset paths for the skills to retrieve. Example: ["/Game/Skills/MySkill.MySkill_C"].
Returns: object mapping SkillPath → { "instructions": string }

​
CreateSkill

Creates a new AgentSkill asset in the project.
​
folderPath
string
required
Content Browser path where the asset will be created. Example: "/Game/Skills/".
​
assetName
string
required
Name of the new asset (without path or extension). Example: "MySkill".
​
description
string
required
A brief, human-readable summary of what the skill does.
​
details
object
required
An object with a single key instructions (string) containing the full skill instructions.
Returns: string — the full asset path of the created skill class, or an empty string if creation failed. Example 
{
  "name": "call_tool",
  "arguments": {
    "toolset_name": "ToolsetRegistry.AgentSkillToolset",
    "tool_name": "CreateSkill",
    "arguments": {
      "folderPath": "/Game/Skills/",
      "assetName": "LightingSkill",
      "description": "Rules for placing dynamic lighting in a scene.",
      "details": {
        "instructions": "Always group lights in the Outliner. Prefer PointLight over SpotLight for fill lighting."
      }
    }
  }
}

​
UpdateSkill

Modifies the description or instructions of an existing AgentSkill asset.
​
skillPath
string
required
Full asset path including the _C suffix. Example: "/Game/Skills/MySkill.MySkill_C".
​
description
string
required
Updated brief description of the skill.
​
details
object
required
Object with a single key instructions (string) with the updated full skill text.
Returns: boolean — true if the update succeeded, false otherwise.

​
Tool Argument Format Notes

​
UObject / UClass references — refPath

Tools that accept a class or asset reference expect an object with a refPath key, not a plain string: 
// ❌ Wrong — plain string
{ "actor_type": "/Script/Engine.PointLight" }

// ✅ Correct — refPath object
{ "actor_type": { "refPath": "/Script/Engine.PointLight" } }

​
Transform format — xform

Tools that accept a world transform use the ToolsetTransform schema. Key names are location and scale — not translation or scale3d: 
{
  "xform": {
    "location": { "x": 0, "y": 0, "z": 300 },
    "rotation": { "pitch": 0, "yaw": 0, "roll": 0 },
    "scale":    { "x": 1, "y": 1, "z": 1 }
  }
}
Using translation or scale3d instead of location / scale will cause the tool call to fail silently or apply identity values. Always match the key names exactly as shown above.

Build docs developers (and LLMs) love

Get started for freeTalk to us