Unity + Koog: 從 Kotlin 代理程式驅動您的遊戲

在 GitHub 上開啟 下載 .ipynb

本筆記本將引導您使用 Model Context Protocol (MCP) 和 Koog 建立一個精通 Unity 的 AI 代理程式。我們將連線到 Unity MCP 伺服器、探索工具、使用 LLM 進行規劃，並對您開啟的場景執行動作。

先決條件

• 一個已安裝 Unity-MCP 伺服器外掛程式的 Unity 專案
• JDK 17+
• OPENAI_API_KEY 環境變數中的 OpenAI API 金鑰

%useLatestDescriptors
%use koog

lateinit var process: Process

1) 提供您的 OpenAI API 金鑰

我們從 OPENAI_API_KEY 環境變數讀取 API 金鑰，這樣您可以將機密資訊保存在筆記本之外。

val token = System.getenv("OPENAI_API_KEY") ?: error("OPENAI_API_KEY environment variable not set")
val executor = simpleOpenAIExecutor(token)

2) 配置 Unity 代理程式

我們為 Unity 定義一個簡潔的系統提示和代理程式設定。

val agentConfig = AIAgentConfig(
    prompt = prompt("cook_agent_system_prompt") {
        system {
            "You are a Unity assistant. You can execute different tasks by interacting with tools from the Unity engine."
        }
    },
    model = OpenAIModels.Chat.GPT4o,
    maxAgentIterations = 1000
)

3) 啟動 Unity MCP 伺服器

我們將從您的 Unity 專案目錄啟動 Unity MCP 伺服器，並透過 stdio 連線。

// https://github.com/IvanMurzak/Unity-MCP
val pathToUnityProject = "path/to/unity/project"
val process = ProcessBuilder(
    "$pathToUnityProject/com.ivanmurzak.unity.mcp.server/bin~/Release/net9.0/com.IvanMurzak.Unity.MCP.Server",
    "60606"
).start()

4) 從 Koog 連線並執行代理程式

我們從 Unity MCP 伺服器探索工具、建立一個小型「計劃優先」策略，並執行一個僅使用工具來修改您開啟場景的代理程式。

import kotlinx.coroutines.runBlocking

runBlocking {
    // Create the ToolRegistry with tools from the MCP server
    val toolRegistry = McpToolRegistryProvider.fromTransport(
        transport = McpToolRegistryProvider.defaultStdioTransport(process)
    ) + ToolRegistry {
        tool(ProvideStringSubgraphResult)
    }

    toolRegistry.tools.forEach {
        println(it.name)
        println(it.descriptor)
    }

    val strategy = strategy<String, String>("unity_interaction") {
        val nodePlanIngredients by nodeLLMRequest(allowToolCalls = false)
        val interactionWithUnity by subgraphWithTask<String>(
            // work with plan
            tools = toolRegistry.tools,
        ) { input ->
            "Start interacting with Unity according to the plan: $input"
        }

        edge(
            nodeStart forwardTo nodePlanIngredients transformed {
                "Create detailed plan for " + agentInput + "" +
                    "using the following tools: ${toolRegistry.tools.joinToString("
") {
                        it.name + "
description:" + it.descriptor
                    }}"
            }
        )
        edge(nodePlanIngredients forwardTo interactionWithUnity onAssistantMessage { true })
        edge(interactionWithUnity forwardTo nodeFinish transformed { it.result })
    }

    val agent = AIAgent(
        promptExecutor = executor,
        strategy = strategy,
        agentConfig = agentConfig,
        toolRegistry = toolRegistry,
        installFeatures = {
            install(Tracing)

            install(EventHandler) {
                onBeforeAgentStarted { eventContext ->
                    println("OnBeforeAgentStarted first (strategy: ${strategy.name})")
                }

                onBeforeAgentStarted { eventContext ->
                    println("OnBeforeAgentStarted second (strategy: ${strategy.name})")
                }

                onAgentFinished { eventContext ->
                    println(
                        "OnAgentFinished (agent id: ${eventContext.agentId}, result: ${eventContext.result})"
                    )
                }
            }
        }
    )

    val result = agent.run(
        " extend current opened scene for the towerdefence game. " +
            "Add more placements for the towers, change the path for the enemies"
    )

    result
}

5) 關閉 MCP 處理程序

在執行結束時，務必清理外部的 Unity MCP 伺服器處理程序。

// Shutdown the Unity MCP process
process.destroy()