{"id":29409993,"url":"https://github.com/ddpie/unity-strands-agent","last_synced_at":"2026-05-19T06:05:27.249Z","repository":{"id":303185457,"uuid":"1014605072","full_name":"ddpie/unity-strands-agent","owner":"ddpie","description":"基于 AWS Strands Agent SDK 的 Unity 开发 AI 助手，集成到 Unity 编辑器中提供智能代码生成、项目优化和实时问答。支持多种 AI 模型、内置工具和MCP，显著提升开发效率。","archived":false,"fork":false,"pushed_at":"2025-07-06T08:18:00.000Z","size":1023,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-07-06T08:19:22.335Z","etag":null,"topics":["ai","gendev","strands-agent-sdk","unity","unity-plugin"],"latest_commit_sha":null,"homepage":"","language":"C#","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ddpie.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-07-06T03:47:12.000Z","updated_at":"2025-07-06T08:18:04.000Z","dependencies_parsed_at":"2025-07-06T08:29:40.209Z","dependency_job_id":null,"html_url":"https://github.com/ddpie/unity-strands-agent","commit_stats":null,"previous_names":["ddpie/unity-strands-agent"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/ddpie/unity-strands-agent","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ddpie%2Funity-strands-agent","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ddpie%2Funity-strands-agent/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ddpie%2Funity-strands-agent/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ddpie%2Funity-strands-agent/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ddpie","download_url":"https://codeload.github.com/ddpie/unity-strands-agent/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ddpie%2Funity-strands-agent/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33204090,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-18T09:27:30.708Z","status":"online","status_checked_at":"2026-05-19T02:00:06.763Z","response_time":58,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","gendev","strands-agent-sdk","unity","unity-plugin"],"created_at":"2025-07-11T05:05:29.391Z","updated_at":"2026-05-19T06:05:27.243Z","avatar_url":"https://github.com/ddpie.png","language":"C#","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Unity Strands Agent\n\n基于 AWS [Strands Agents SDK](https://strandsagents.com/latest/) 的 Unity 编辑器 AI 插件，在 Unity 编辑器中直接与 AI 对话，获得问题解答、学习指导和开发建议。\n\n**Language / 语言**：[中文](#中文版本) | [English](#english-version)\n\n---\n\n## 中文版本\n\nUnity Strands Agent 是一个友好易用的 AI 助手，将 AWS 开源的 [Strands Agents SDK](https://strandsagents.com/latest/) 无缝集成到 Unity 编辑器中。该插件内置了对 Unity 项目的深度理解，能够通过自然语言对话帮助您学习 Unity、解决问题、理解项目结构。**通过支持 Model Context Protocol (MCP)，本插件可以与 [mcp-unity](https://github.com/CoderGamester/mcp-unity) 或其他 Unity MCP 插件灵活搭配**，让您轻松上手 Unity 开发，大幅降低学习门槛。\n\n## 核心实现\n\n基于 [Strands Agents SDK](https://strandsagents.com/latest/) 开发智能助手极其简单，核心代码展示了其优雅的设计：\n\n```python\nfrom strands import Agent\nfrom unity_tools import get_unity_tools\n\nclass UnityAgent:\n    def __init__(self):\n        # 初始化 MCP 管理器\n        from mcp_manager import MCPManager\n        self.mcp_manager = MCPManager()\n        \n        # 从配置文件加载 MCP 服务器并建立连接\n        # 配置示例 (Assets/UnityAIAgent/mcp_config.json):\n        # {\n        #   \"mcpServers\": {\n        #     \"mcp-unity\": {\n        #       \"command\": \"/usr/local/bin/node\",\n        #       \"args\": [\"path/to/mcp-unity/Server/build/index.js\"],\n        #       \"env\": {\"UNITY_PORT\": \"8090\"}\n        #     }\n        #   }\n        # }\n        mcp_tools = self.mcp_manager.load_mcp_tools()\n        \n        # 获取 Unity 开发工具集（包含内置工具和 MCP 工具）\n        unity_tools = get_unity_tools(include_mcp=True, agent_instance=self)\n        \n        # 创建强大的 AI Agent，配备完整工具集\n        self.agent = Agent(\n            system_prompt=UNITY_SYSTEM_PROMPT, \n            tools=unity_tools  # 21+内置工具 + 动态加载的 MCP 工具\n        )\n    \n    def _load_mcp_tools(self):\n        # 由 get_unity_tools 调用，返回已加载的 MCP 工具\n        return self.mcp_manager._mcp_tools\n    \n    def process_message(self, message: str):\n        # 处理用户消息，支持 Unity 原生操作\n        return self.agent(message)\n    \n    def cleanup(self):\n        # 清理 MCP 连接和资源\n        self.mcp_manager.cleanup()\n```\n\n通过以上代码可以看到：\n- **MCP 配置加载**：从 JSON 配置文件读取 MCP 服务器设置\n- **动态工具发现**：MCP 服务器启动后自动发现并加载可用工具\n- **工具集整合**：将内置工具与 MCP 工具无缝整合\n- **资源管理**：提供清理机制确保连接正确关闭\n\n该插件基于模块化架构设计，通过 MCP 协议轻松扩展工具集，让 AI Agent 能够直接操作 Unity 编辑器。\n\n## 主要特性\n\n### Unity 专业化\n- **项目结构理解**：智能识别 Unity 项目结构，提供清晰易懂的解释和指导\n- **学习助手**：耐心解答 Unity 概念、工作流程和最佳实践，适合各种经验水平\n- **问题解决**：快速诊断常见问题，提供详细的解决步骤和方案\n- **编辑器操作**：通过 MCP 协议与 Unity MCP 插件（如 mcp-unity）搭配，实现场景操作、资源管理等直观交互\n\n### 简单易用\n- **自然语言交互**：用日常语言描述您的需求，无需学习复杂的技术术语\n- **一键安装配置**：自动检测环境、安装依赖，几分钟即可开始使用\n- **智能引导**：根据您的问题提供循序渐进的学习建议和操作指导\n- **MCP 协议支持**：通过 Model Context Protocol 灵活扩展能力，可与 [mcp-unity](https://github.com/CoderGamester/mcp-unity/blob/main/README_zh-CN.md) 或其他 Unity MCP 插件无缝集成\n\n### 智能助手\n- **24/7 在线支持**：随时为您解答 Unity 相关问题，从基础概念到高级技巧\n- **个性化学习**：根据您的项目和需求，提供定制化的学习路径和建议\n- **持续改进**：通过与您的对话不断学习，提供越来越精准的帮助\n\n### AI 模型支持\n- **多模型兼容**：[Strands Agents SDK](https://strandsagents.com/latest/) 支持 Claude、GPT、Llama 等主流 AI 模型\n- **当前支持**：本插件当前版本仅支持 Bedrock US Oregon (us-west-2) 区域的 Claude 3.7 Sonnet\n- **智能记忆**：记住您的学习进度和项目偏好，提供个性化学习建议\n\n## 系统架构\n\n```mermaid\n%%{init: {'theme':'neutral'}}%%\ngraph TD\n    subgraph Unity[\"Unity 编辑器\"]\n        UI[Unity AI Agent Window\u003cbr/\u003e用户界面]\n        PB[Python Bridge\u003cbr/\u003ePython桥接层]\n    end\n    \n    subgraph AgentPython[\"智能 Agent 系统\"]\n        AC[Agent Core\u003cbr/\u003eAgent 核心]\n        UA[Unity Agent\u003cbr/\u003eUnity 专用 Agent]\n        SDK[Strands Agents SDK\u003cbr/\u003e核心AI框架]\n    end\n    \n    subgraph ModelServices[\"AI 模型服务\"]\n        AWS[Amazon Bedrock\u003cbr/\u003e✓ 插件已支持]\n        ANTHROPIC[Anthropic Claude\u003cbr/\u003eSDK支持]\n        OPENAI[OpenAI GPT\u003cbr/\u003eSDK支持]\n        OLLAMA[Ollama 本地模型\u003cbr/\u003eSDK支持]\n        OTHERS[LiteLLM / Llama 等\u003cbr/\u003eSDK支持]\n    end\n    \n    subgraph MCPServices[\"扩展服务生态\"]\n        MCP[MCP Servers\u003cbr/\u003e扩展服务]\n    end\n    \n    %% Connections\n    UI --\u003e|用户交互| PB\n    PB --\u003e|Python.NET| AC\n    AC --\u003e|创建| UA\n    UA ==\u003e|依赖| SDK\n    SDK ==\u003e|调用| AWS\n    SDK -.-\u003e|支持但未集成| ANTHROPIC\n    SDK -.-\u003e|支持但未集成| OPENAI\n    SDK -.-\u003e|支持但未集成| OLLAMA\n    SDK -.-\u003e|支持但未集成| OTHERS\n    SDK --\u003e|调用| MCP\n    \n    %% Subgraph styles\n    style Unity fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    style AgentPython fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    style ModelServices fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    style MCPServices fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    \n    %% Node styles - Blue for supported features\n    style UI fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style PB fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style AC fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style UA fill:#1976D2,stroke:#0D47A1,stroke-width:3px,color:#FFF\n    style AWS fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style MCP fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    \n    %% Core SDK - Red for emphasis\n    style SDK fill:#D32F2F,stroke:#B71C1C,stroke-width:4px,color:#FFF\n    \n    %% SDK Only services - Gray\n    style ANTHROPIC fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n    style OPENAI fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n    style OLLAMA fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n    style OTHERS fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n```\n\n## 快速开始\n\n### 3分钟快速上手\n\n1. **安装插件**\n   ```\n   Unity Package Manager → Add package from git URL\n   输入：https://github.com/ddpie/unity-strands-agent.git\n   ```\n\n2. **自动配置**\n   ```\n   Window → Unity Strands Agent → Settings → Environment Setup → Start Setup\n   等待自动安装完成（无需任何手动配置）\n   ```\n\n3. **开始聊天**\n   ```\n   点击 `Open AI Assistant`，像聊天一样问问题即可\n   例如：\"Unity中的GameObject是什么？\"\n   ```\n\n**系统要求**：Unity 2022.3+ | Python 3.11 | macOS | AWS credentials\n\n**视频教程**：[Unity Strands Agent 功能演示](https://www.bilibili.com/video/BV1jgGJzhEZs/)\n\n![Unity Strands Agent 中文界面](Assets/Resources/unity_strands_agent_demo.png)\n\n## 详细配置\n\n### 系统要求\n\n- Unity 2022.3 LTS 或更高版本\n- Python 3.11（插件会自动检测和配置）\n- AWS credentials 配置（默认使用 Amazon Bedrock 服务，默认模型为 US Oregon (us-west-2) 区域的 Claude 3.7 Sonnet）\n- **仅支持 macOS 系统**\n\n**重要**：\n- 本插件目前仅适配 Python 3.11，暂不支持其他 Python 版本。在 macOS 上可通过 `brew install python@3.11` 安装。\n- 环境初始化将在点击 `Environment Setup` 的 `Start Setup` 按钮时进行，请耐心等待安装过程完成。\n\n### 安装步骤\n\n通过 Unity Package Manager 安装：\n\n1. 在 Unity 中打开 Package Manager\n2. 点击 \"+\" 按钮\n3. 选择 \"Add package from git URL\"\n4. 输入：`https://github.com/ddpie/unity-strands-agent.git`\n5. 点击 \"Add\" 并等待 Unity 自动下载和导入\n\n### 配置步骤\n\n1. 在 Unity 中选择菜单：`Window → Unity Strands Agent`\n2. 点击 `Settings` 标签页，然后点击 `Environment Setup`\n3. 点击 `Start Setup` 按钮开始自动安装\n4. 系统将自动完成环境检测、虚拟环境创建、依赖安装等步骤\n5. 配置完成后，点击 `Open AI Assistant` 开始使用\n\n### 环境配置\n\n插件会自动设置以下环境变量：\n\n- `PROJECT_ROOT_PATH`：项目根目录路径\n- `STRANDS_TOOLS_PATH`：Strands工具路径\n\n配置文件保存在 `Assets/UnityAIAgent/PathConfiguration.asset` 中，会自动加载。插件使用 AWS credentials 配置文件访问 Bedrock 服务。\n\n**MCP 服务器配置**：本插件支持通过 MCP 协议与各种 Unity MCP 插件搭配使用。在 Unity 编辑器界面的 Settings → MCP Configuration 中以 JSON 格式添加服务器配置。建议使用绝对路径配置 `command` 字段，例如配置 mcp-unity 插件：\n```json\n{\n  \"mcpServers\": {\n    \"mcp-unity\": {\n      \"command\": \"/usr/local/bin/node\",\n      \"env\": {\n        \"UNITY_PORT\": \"8090\"\n      },\n      \"args\": [\n        \"/path/to/your/unity-project/Library/PackageCache/com.gamelovers.mcp-unity@latest/Server/build/index.js\"\n      ]\n    }\n  }\n}\n```\n\n您可以根据需求选择和配置不同的 Unity MCP 插件。关于 mcp-unity 插件的详细安装和使用说明，请参考：[mcp-unity 插件文档](https://github.com/CoderGamester/mcp-unity/blob/main/README_zh-CN.md)\n\n### 故障排除\n\n**菜单项不显示**\n- 检查Unity控制台是否有编译错误\n- 确认包已正确下载到Library/PackageCache目录\n- 重启Unity编辑器\n\n**Python 3.11 环境问题**\n- 确认系统已安装 Python 3.11：`python3.11 --version`\n- 在 macOS 上安装：`brew install python@3.11`\n- 重新运行 `Environment Setup` 进行自动配置\n\n**找不到agent_core模块**\n- 运行 `Environment Setup` 重新安装Python依赖\n- 检查Library/PackageCache中是否有com.ddpie.unity-strands-agent包\n- 确认虚拟环境在项目Python目录中正确创建\n\n如有问题，请检查Unity控制台的错误日志，并在设置页面点击 `验证配置` 查看当前配置状态。\n\n### 支持和反馈\n\n- **文档**：[Strands Agents SDK 官方文档](https://strandsagents.com/latest/)\n- **问题反馈**：[GitHub Issues](https://github.com/ddpie/unity-strands-agent/issues)\n\n### 参考资料\n\n#### Strands Agents SDK\n- **官方文档**：https://strandsagents.com/latest/\n- **Python SDK**：https://github.com/strands-agents/sdk-python\n- **示例代码**：https://github.com/strands-agents/samples\n- **工具包**：https://github.com/strands-agents/tools\n- **Agent Builder**：https://github.com/strands-agents/agent-builder\n\n#### 技术实现\n- **Python.NET**：https://github.com/pythonnet/pythonnet\n- **Model Context Protocol (MCP)**：https://modelcontextprotocol.io/\n- **Unity Editor 扩展**：https://docs.unity3d.com/Manual/ExtendingTheEditor.html\n\n#### 相关项目\n- **mcp-unity 插件**：https://github.com/CoderGamester/mcp-unity\n- **PythonForUnity**：https://github.com/Maesla/PythonForUnity\n\n---\n\n## English Version\n\nUnity Strands Agent is a friendly and accessible AI assistant that seamlessly integrates AWS's open-source [Strands Agents SDK](https://strandsagents.com/latest/) into the Unity Editor. Built with deep understanding of Unity project structures and workflows, this plugin helps you learn Unity, solve problems, and understand project organization through natural language conversations. **Through support for the Model Context Protocol (MCP), it can be flexibly paired with [mcp-unity](https://github.com/CoderGamester/mcp-unity) or other Unity MCP plugins** to make Unity development approachable and significantly lower the learning curve.\n\n## Core Implementation\n\nBuilding intelligent assistants with the [Strands Agents SDK](https://strandsagents.com/latest/) is remarkably simple. The core implementation showcases its elegant design:\n\n```python\nfrom strands import Agent\nfrom unity_tools import get_unity_tools\n\nclass UnityAgent:\n    def __init__(self):\n        # Initialize MCP manager\n        from mcp_manager import MCPManager\n        self.mcp_manager = MCPManager()\n        \n        # Load MCP servers from config and establish connections\n        # Config example (Assets/UnityAIAgent/mcp_config.json):\n        # {\n        #   \"mcpServers\": {\n        #     \"mcp-unity\": {\n        #       \"command\": \"/usr/local/bin/node\",\n        #       \"args\": [\"path/to/mcp-unity/Server/build/index.js\"],\n        #       \"env\": {\"UNITY_PORT\": \"8090\"}\n        #     }\n        #   }\n        # }\n        mcp_tools = self.mcp_manager.load_mcp_tools()\n        \n        # Load Unity development toolset (includes built-in and MCP tools)\n        unity_tools = get_unity_tools(include_mcp=True, agent_instance=self)\n        \n        # Create a powerful AI Agent with complete toolset\n        self.agent = Agent(\n            system_prompt=UNITY_SYSTEM_PROMPT, \n            tools=unity_tools  # 21+ built-in tools + dynamically loaded MCP tools\n        )\n    \n    def _load_mcp_tools(self):\n        # Called by get_unity_tools to retrieve loaded MCP tools\n        return self.mcp_manager._mcp_tools\n    \n    def process_message(self, message: str):\n        # Process user messages with Unity native operations support\n        return self.agent(message)\n    \n    def cleanup(self):\n        # Cleanup MCP connections and resources\n        self.mcp_manager.cleanup()\n```\n\nThe code above demonstrates:\n- **MCP Configuration Loading**: Reads MCP server settings from JSON configuration files\n- **Dynamic Tool Discovery**: Automatically discovers and loads available tools after MCP server startup\n- **Tool Integration**: Seamlessly combines built-in tools with MCP tools\n- **Resource Management**: Provides cleanup mechanisms to ensure connections are properly closed\n\nBuilt on a modular architecture, this plugin leverages the MCP protocol to extend the toolset, enabling the AI Agent to directly operate the Unity Editor.\n\n## Key Features\n\n### Unity Specialization\n- **Project Understanding**: Intelligently analyzes your Unity project structure and provides clear, easy-to-understand explanations\n- **Learning Companion**: Patiently explains Unity concepts, workflows, and best practices suitable for all experience levels\n- **Problem Solver**: Quickly diagnoses common issues and provides detailed step-by-step solutions\n- **Editor Operations**: Through MCP protocol integration with Unity MCP plugins (such as mcp-unity), enables intuitive scene operations, asset management, and more\n\n### Simple and Accessible\n- **Natural Language Interaction**: Describe your needs in everyday language without learning complex technical terminology\n- **One-Click Setup**: Automatically detects environment, installs dependencies, ready to use in minutes\n- **Intelligent Guidance**: Provides step-by-step learning suggestions and operation guidance based on your questions\n- **MCP Protocol Support**: Flexibly extend capabilities through the Model Context Protocol, seamlessly integrating with [mcp-unity](https://github.com/CoderGamester/mcp-unity/blob/main/README.md) or other Unity MCP plugins\n\n### Intelligent Assistant\n- **24/7 Support**: Always available to answer your Unity questions, from basic concepts to advanced techniques\n- **Personalized Learning**: Provides customized learning paths and suggestions based on your projects and needs\n- **Continuous Improvement**: Learns from your conversations to provide increasingly accurate and helpful assistance\n\n### AI Model Support\n- **Multi-Model Compatibility**: The [Strands Agents SDK](https://strandsagents.com/latest/) supports Claude, GPT, Llama, and other leading AI models\n- **Current Support**: This plugin currently only supports Claude 3.7 Sonnet in Bedrock US Oregon (us-west-2) region\n- **Intelligent Memory**: Automatically remembers your project patterns, coding style, and preferences across sessions\n\n## System Architecture\n\n```mermaid\n%%{init: {'theme':'neutral'}}%%\ngraph TD\n    subgraph Unity[\"Unity Editor\"]\n        UI[Unity AI Agent Window\u003cbr/\u003eUser Interface]\n        PB[Python Bridge\u003cbr/\u003eIntegration Bridge]\n    end\n    \n    subgraph AgentPython[\"AI Agent System\"]\n        AC[Agent Core\u003cbr/\u003eCore Agent System]\n        UA[Unity Agent\u003cbr/\u003eUnity Specialized Agent]\n        SDK[Strands Agents SDK\u003cbr/\u003eCore AI Framework]\n    end\n    \n    subgraph ModelServices[\"AI Model Services\"]\n        AWS[Amazon Bedrock\u003cbr/\u003e✓ Plugin Ready]\n        ANTHROPIC[Anthropic Claude\u003cbr/\u003eSDK Only]\n        OPENAI[OpenAI GPT\u003cbr/\u003eSDK Only]\n        OLLAMA[Ollama Local Models\u003cbr/\u003eSDK Only]\n        OTHERS[LiteLLM / Llama etc\u003cbr/\u003eSDK Only]\n    end\n    \n    subgraph MCPServices[\"Extension Ecosystem\"]\n        MCP[MCP Servers\u003cbr/\u003eExtension Services]\n    end\n    \n    %% Connections\n    UI --\u003e|User Interaction| PB\n    PB --\u003e|Python.NET| AC\n    AC --\u003e|Creates| UA\n    UA ==\u003e|Depends on| SDK\n    SDK ==\u003e|Calls| AWS\n    SDK -.-\u003e|Supports but not integrated| ANTHROPIC\n    SDK -.-\u003e|Supports but not integrated| OPENAI\n    SDK -.-\u003e|Supports but not integrated| OLLAMA\n    SDK -.-\u003e|Supports but not integrated| OTHERS\n    SDK --\u003e|Calls| MCP\n    \n    %% Subgraph styles\n    style Unity fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    style AgentPython fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    style ModelServices fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    style MCPServices fill:#E3F2FD,stroke:#1976D2,stroke-width:2px\n    \n    %% Node styles - Blue for supported features\n    style UI fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style PB fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style AC fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style UA fill:#1976D2,stroke:#0D47A1,stroke-width:3px,color:#FFF\n    style AWS fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    style MCP fill:#BBDEFB,stroke:#1976D2,stroke-width:2px\n    \n    %% Core SDK - Red for emphasis\n    style SDK fill:#D32F2F,stroke:#B71C1C,stroke-width:4px,color:#FFF\n    \n    %% SDK Only services - Gray\n    style ANTHROPIC fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n    style OPENAI fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n    style OLLAMA fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n    style OTHERS fill:#EEEEEE,stroke:#9E9E9E,stroke-width:1px\n```\n\n## Quick Start\n\n### 3-Minute Quick Start\n\n1. **Install Plugin**\n   ```\n   Unity Package Manager → Add package from git URL\n   Enter: https://github.com/ddpie/unity-strands-agent.git\n   ```\n\n2. **Auto Setup**\n   ```\n   Window → Unity Strands Agent → Settings → Environment Setup → Start Setup\n   Wait for automatic installation (no manual configuration needed)\n   ```\n\n3. **Start Chatting**\n   ```\n   Click `Open AI Assistant` and ask questions like chatting\n   Example: \"What is a GameObject in Unity?\"\n   ```\n\n**Requirements**: Unity 2022.3+ | Python 3.11 | macOS | AWS credentials\n\n**Video Demo**: [Unity Strands Agent Demo](https://www.bilibili.com/video/BV1jgGJzhEZs/)\n\n![Unity Strands Agent English Interface](Assets/Resources/unity_strands_agent_demo_en.png)\n\n## Detailed Configuration\n\n### System Requirements\n\n- Unity 2022.3 LTS or higher\n- Python 3.11 (plugin automatically detects and configures)\n- AWS credentials configuration (default model provider is Amazon Bedrock with Claude 3.7 Sonnet in the US Oregon (us-west-2) region)\n- **macOS only**\n\n**Important**:\n- This plugin currently only supports Python 3.11 and does not support other Python versions. On macOS, install with `brew install python@3.11`.\n- Environment initialization will occur when you click the `Start Setup` button in `Environment Setup`, please wait for the installation process to complete.\n\n### Installation Steps\n\nThrough Unity Package Manager:\n\n1. Open Package Manager in Unity\n2. Click \"+\" button\n3. Select \"Add package from git URL\"\n4. Enter: `https://github.com/ddpie/unity-strands-agent.git`\n5. Click \"Add\" and wait for Unity to automatically download and import\n\n### Configuration Steps\n\n1. In Unity, select menu: `Window → Unity Strands Agent`\n2. Click `Settings` tab, then click `Environment Setup`\n3. Click `Start Setup` button to begin automatic installation\n4. The system will automatically complete environment detection, virtual environment creation, dependency installation\n5. After configuration completes, click `Open AI Assistant` to start using\n\nMain capabilities include:\n- Clear explanations of Unity concepts and project structure\n- Step-by-step guidance for common Unity tasks and workflows\n- Helpful suggestions for learning Unity development\n- Friendly assistance with Unity-specific questions and challenges\n\n### Environment Configuration\n\nThe plugin automatically sets the following environment variables:\n\n- `PROJECT_ROOT_PATH`: Project root directory path\n- `STRANDS_TOOLS_PATH`: Strands tools path\n\nConfiguration is saved in `Assets/UnityAIAgent/PathConfiguration.asset` and will be automatically loaded. The plugin uses AWS credentials configuration files to access Bedrock services.\n\n**MCP Server Configuration**: This plugin supports pairing with various Unity MCP plugins through the MCP protocol. Configure MCP servers through the Unity editor interface at Settings → MCP Configuration in JSON format. It's recommended to use absolute paths for the `command` field, for example configuring the mcp-unity plugin:\n```json\n{\n  \"mcpServers\": {\n    \"mcp-unity\": {\n      \"command\": \"/usr/local/bin/node\",\n      \"env\": {\n        \"UNITY_PORT\": \"8090\"\n      },\n      \"args\": [\n        \"/path/to/your/unity-project/Library/PackageCache/com.gamelovers.mcp-unity@latest/Server/build/index.js\"\n      ]\n    }\n  }\n}\n```\n\nYou can choose and configure different Unity MCP plugins based on your needs. For detailed installation and usage instructions for the mcp-unity plugin, please refer to: [mcp-unity Plugin Documentation](https://github.com/CoderGamester/mcp-unity/blob/main/README.md)\n\n### Troubleshooting\n\n**Menu Item Not Showing**\n- Check Unity console for compilation errors\n- Confirm package has been properly downloaded to Library/PackageCache directory\n- Restart Unity Editor\n\n**Python 3.11 Environment Issues**\n- Confirm Python 3.11 is installed on system: `python3.11 --version`\n- Install on macOS: `brew install python@3.11`\n- Re-run `Environment Setup` for automatic configuration\n\n**Cannot Find agent_core Module**\n- Run `Environment Setup` to reinstall Python dependencies\n- Check if com.ddpie.unity-strands-agent package exists in Library/PackageCache\n- Confirm virtual environment is correctly created in project Python directory\n\nIf you encounter issues, check Unity console for error logs and click `Validate Configuration` in settings page to view current configuration status.\n\n### Support and Feedback\n\n- **Documentation**: [Strands Agents SDK Official Documentation](https://strandsagents.com/latest/)\n- **Issue Reporting**: [GitHub Issues](https://github.com/ddpie/unity-strands-agent/issues)\n\n### References\n\n#### Strands Agents SDK\n- **Official Documentation**: https://strandsagents.com/latest/\n- **Python SDK**: https://github.com/strands-agents/sdk-python\n- **Sample Code**: https://github.com/strands-agents/samples\n- **Tools Package**: https://github.com/strands-agents/tools\n- **Agent Builder**: https://github.com/strands-agents/agent-builder\n\n#### Technical Implementation\n- **Python.NET**: https://github.com/pythonnet/pythonnet\n- **Model Context Protocol (MCP)**: https://modelcontextprotocol.io/\n- **Unity Editor Extensions**: https://docs.unity3d.com/Manual/ExtendingTheEditor.html\n\n#### Related Projects\n- **mcp-unity Plugin**: https://github.com/CoderGamester/mcp-unity\n- **PythonForUnity**: https://github.com/Maesla/PythonForUnity\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fddpie%2Funity-strands-agent","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fddpie%2Funity-strands-agent","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fddpie%2Funity-strands-agent/lists"}