Unreal Engine Knowledge Graph

by yarnovo

Not rated
GitHub

About

Search concept relationships in the Unreal Engine official documentation using a Neo4j-powered knowledge graph.

Details

Author
yarnovo
Categories
Search, Knowledge Base, Other

Setup

Install Unreal Engine Knowledge Graph in your MCP client (Claude Desktop, Cursor, Windsurf, and others).

Repository: https://github.com/yarnovo/unreal-engine-knowledge-graph-mcp

Follow the installation instructions in the repository README, then restart your MCP client.

Search concept relationships in the Unreal Engine official documentation using a Neo4j-powered knowledge graph.

这个项目提供虚幻引擎官方文档的 MCP(Model Context Protocol)服务器,支持基于Neo4j图数据库的概念关系搜索,帮助开发者发现概念间的学习路径和依赖关系。

- 蓝图系统与C++代码的关系
- 材质编辑器与节点图编程的关系
- Nanite虚拟几何体与高多边形模型的关系

传统的文档搜索只能找到单个概念的信息,无法揭示概念间的学习路径和依赖关系。本项目通过构建知识图谱,让AI能够理解概念间的关联,提供更智能的学习指导。

本项目提供了一个基于Neo4j图数据库的MCP服务器,专门用于虚幻引擎概念关系的智能搜索和发现。通过DeepSeek v3模型提取文档中的概念关系,构建完整的知识图谱。

- 🔗概念关系搜索: 发现任意概念的相关概念和学习路径
- 🧠智能概念发现: 基于图数据库的深度关系挖掘
- 🔍概念名称搜索: 模糊搜索概念名称,支持中英文双语查询
- 📊关系统计信息: 获取概念的关系数量统计,按重要性排序

{ "mcpServers": { "unreal-engine-knowledge-graph-mcp": { "command": "npx", "args": [ "-y", "unreal-engine-knowledge-graph-mcp" ], "env": { "NEO4J_URI": "bolt://localhost:7687", "NEO4J_USER": "neo4j", "NEO4J_PASSWORD": "password123" } } } }
{ "servers": { "unreal-engine-knowledge-graph-mcp": { "type": "stdio", "command": "npx", "args": [ "-y", "unreal-engine-knowledge-graph-mcp" ], "env": { "NEO4J_URI": "bolt://localhost:7687", "NEO4J_USER": "neo4j", "NEO4J_PASSWORD": "password123" } } } }

- 🎯概念学习扩展: "我想学习蓝图系统,相关的概念还有哪些?"
- 🔍技术关联探索: "虚幻引擎包含哪些核心功能模块?"
- 🧭学习路径规划: "从材质编辑器出发,我还需要了解什么?"

帮我搜索"蓝图系统"和"Blueprint System"的相关概念,我想了解它与其他功能的关系 查找"虚幻引擎"和"Unreal Engine"包含哪些核心功能 搜索"材质编辑器"和"Material Editor"相关的学习内容

- concept(必需): 要查询的概念名称(中英文双语)

- cn(必需): 中文概念名称
- en(必需): 英文概念名称

{ "searchTerms": { "cn": "虚幻引擎", "en": "Unreal Engine" }, "concept": "虚幻引擎 / Unreal Engine", "found": true, "totalRelations": 12, "relatedConcepts": [ { "concept": "蓝图系统", "predicate": "包含", "context": "虚幻引擎的可视化脚本编程系统", "direction": "outgoing" } ], "limit": 20 }

- 🔍快速查找概念: "我记得有个关于'粒子'的功能,叫什么名字来着?"
- 📝概念名称确认: "虚幻引擎中2D相关的功能都有哪些?"
- 🎯关键词探索: "搜索包含'编辑器'的所有概念"

搜索包含"粒子"和"Particle"的所有概念 查找与"2D"相关的功能 搜索"编辑器"和"Editor"相关的工具

- searchTerm(必需): 搜索关键词(中英文双语)

- cn(必需): 中文搜索关键词
- en(必需): 英文搜索关键词

{ "searchTerms": { "cn": "蓝图", "en": "Blueprint" }, "concepts": ["蓝图系统", "蓝图编辑器", "Blueprint System", "Blueprint Editor"], "count": 4, "limit": 10 }

- 📋核心概念优先浏览: "虚幻引擎知识图谱中最重要的概念有哪些?"
- 🎯学习计划制定: "我想按重要性顺序学习,哪些是核心概念?"
- 📊概念关系分析: "这些概念分别有多少关联,哪些最核心?"

显示最重要的虚幻引擎概念,按关系数量排序 列出前50个核心概念,我想了解哪些最重要 获取概念列表及其关系统计,帮我制定学习计划
{ "concepts": [ { "concept": "蓝图系统", "relationCount": 25, "incomingCount": 12, "outgoingCount": 13 }, { "concept": "虚幻引擎", "relationCount": 20, "incomingCount": 8, "outgoingCount": 12 } ], "count": 2, "limit": 100, "note": "概念按关系数量从大到小排序,包含入度、出度和总关系数统计" }

- 文档处理: 读取Markdown文档,使用DeepSeek v3提取概念关系
- 知识图谱: 基于Neo4j存储概念和关系数据
- MCP服务: 提供标准化的概念关系查询接口

Markdown文档 → DeepSeek v3分析 → 概念关系提取 → Neo4j图数据库 → MCP工具查询

- Node.js >= 18.0.0
- Docker (用于运行Neo4j)
- DeepSeek API密钥

git clone https://github.com/your-username/unreal-engine-knowledge-graph-mcp.git cd unreal-engine-knowledge-graph-mcp
# 复制环境变量模板 cp .env.example .env # 编辑.env文件,添加DeepSeek API密钥 DEEPSEEK_API_KEY=your_deepseek_api_key_here
# 启动Docker容器 docker-compose up -d # 等待数据库启动完成 npm run test-connection
# 测试模式:只处理一个文档文件 npm run extract-triplets:test-mode
{ "filename": "文档名称", "sourceFile": "源文件路径", "triples": [ { "subject": "主体概念", "predicate": "关系谓词", "object": "客体概念", "context": "上下文说明", "direction": "bidirectional" } ], "timestamp": "创建时间戳" }

- subject: 主体概念名称(知识三元组的主语)
- predicate: 关系谓词(如:包含、支持、依赖、关联等)
- object: 客体概念名称(知识三元组的宾语)
- context: 关系的上下文说明,帮助理解关系的具体含义
- direction: 关系方向性

- "unidirectional": 单向关系(主体→客体,但客体不一定→主体)
- "bidirectional": 双向关系(主体↔客体,相互关联)

- 0.9-1.0: 明确的技术关系,文档中有直接、清晰的说明
- 0.7-0.9: 较为明确的关系,基于上下文推断但证据充分
- 0.5-0.7: 中等置信度,关系存在但需要一定推理
- 0.3-0.5: 较弱的关系,主要基于语义相似性
- 0.1-0.3: 非常弱的关系,仅基于概念共现

├── scripts/ # 脚本文件 │ ├── extract-triplets.ts # 知识三元组提取 │ ├── import-to-neo4j.ts # 数据导入Neo4j │ └── test-connection.ts # 数据库连接测试 ├── bin/ # 源代码 │ ├── index.ts # MCP服务器实现 │ └── neo4j-search.ts # Neo4j搜索引擎 ├── sources/ # 数据文件 │ ├── docs/ # Markdown文档 │ └── triplets/ # 知识三元组JSON文件 ├── tests/ # 测试文件 │ └── mcp-client.test.ts # MCP客户端测试 ├── docker-compose.yml # Neo4j Docker配置 └── package.json # 项目配置
# 构建项目 npm run build # 测试数据库连接 npm run test-connection # 生成演示数据 npm run extract-triplets:test-mode # 提取知识三元组(需要DeepSeek API) npm run extract-triplets # 导入数据到Neo4j npm run import-to-neo4j # 清空数据库并重新导入 npm run import-to-neo4j -- --clear # 运行测试 npm test

- 将Markdown文档放入sources/docs/目录
- 运行概念关系提取:npm run extract-triplets
- 导入到Neo4j:npm run import-to-neo4j

{ "filename": "custom-triples", "sourceFile": "custom/triples.md", "triples": [ { "subject": "自定义概念A", "predicate": "关联", "object": "自定义概念B", "context": "这是一个自定义的知识三元组", "direction": "bidirectional" } ], "timestamp": "2025-01-12T10:30:15.387Z" }

- Node.js: 运行环境
- TypeScript: 类型安全的开发语言
- MCP SDK: Model Context Protocol 实现
- Neo4j: 图数据库
- Docker: 容器化部署

- DeepSeek v3: 概念关系提取
- OpenAI SDK: API调用接口

- Vitest: 单元测试框架
- tsx: TypeScript执行器
- Zod: 参数验证

# 检查Docker容器状态 docker-compose ps # 查看Neo4j日志 docker-compose logs neo4j # 重启Neo4j容器 docker-compose restart neo4j

Search global news using natural language. Webz.io News Search API returns the most relevant articles and content, with filters for source, country, language, date, sentiment, and category.

Search and retrieve documentation for the Panda3D game engine.

Art provenance intelligence: 282,731-node knowledge graph over Getty Provenance Index, Knoedler stock books, and wartime-restitution records — every answer cited, custody gaps named. Free keyless remote: https://provenio.art/api/mcp

Search and retrieve Unity documentation and API references locally with smart version handling.

Remote MCP for commodities research — knowledge graph, COT positioning, natural gas storage/weather, futures quotes, and private stories/journals. Requires Arc Research Starter + bearer token.

Persistent shared memory for AI agents. Hybrid search (pgvector + tsvector), knowledge graph, cognitive scoring - 97.2% Recall@10 on LongMemEval

Stores conversations, connects ideas, remembers context: a knowledge graph for your AI assistants

Persistent coding memory for AI assistants — MCP stdio + remote URL, hybrid search, knowledge graph, token-efficient context. Self-host on Cloudflare D1 or Postgres.

Fetch, convert, and search AWS documentation pages, with recommendations for related content.

Search Stack Overflow and Stack Exchange from Claude, Cursor, and other AI agents through an MCP server.

Quran-focused MCP server for ayah translation, tafsir, mutashabihat lookups, recitation playlists, and prayer times.

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.