mcp-dalamud
Exposes the Dalamud plugin API documentation to AI assistants, enabling them to answer questions about services, types, and methods for writing FFXIV Dalamud plugins.
README
mcp-dalamud
An MCP server that exposes the Dalamud plugin API documentation to AI assistants. Compatible with any MCP-capable client (Claude Desktop, Cursor, Windsurf, VS Code with Copilot, etc.). Useful when writing FFXIV Dalamud plugins — ask your assistant about services, types, and methods without leaving your editor.
Requirements
- Node.js 18+
- Any MCP-compatible AI assistant
Setup
npm install
npm run build
Then register the server with your MCP client. The server communicates over stdio and takes no environment variables.
Claude Desktop (%APPDATA%\Claude\claude_desktop_config.json on Windows, ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"dalamud-api": {
"command": "/path/to/node",
"args": ["/path/to/mcp-dalamud/dist/index.js"]
}
}
}
Cursor / Windsurf / other clients: refer to your client's MCP documentation — the server entry is the same command + args pair above.
Restart your client after editing the config. On first launch the server crawls dalamud.dev/api and writes a local cache to cache/dalamud-api.json (~96 namespaces, ~572 types). Subsequent starts load from disk instantly.
Tools
| Tool | Description |
|---|---|
list_namespaces |
List all Dalamud API namespaces |
get_namespace |
List all types in a namespace, grouped by kind |
get_type |
Full docs for a type — properties, methods, events, declarations |
search |
Keyword search across all types and summaries; optional kind filter (e.g. "enum", "interface") |
list_enums |
Browse all enum types, optionally filtered by namespace or keyword |
list_services |
List injectable services from Dalamud.Plugin.Services |
find_events |
Find event arg types and delegate subscription points across the API |
search_members |
Search properties/methods/events across already-loaded types (no network) |
get_member |
Full docs for a specific member on a type |
health |
Cache status: build time, Dalamud version, member coverage |
refresh_cache |
Re-crawl dalamud.dev and rebuild the cache |
Example prompts
- "What services are available in Dalamud.Plugin.Services?"
- "Show me IClientState — what properties does it have?"
- "Search for anything related to inventory"
- "What enums exist for inventory slot types?"
- "How do I subscribe to framework events in Dalamud?"
- "How do I register a slash command in Dalamud?"
These work with any MCP-capable assistant that supports tool use.
How it works
Startup: loads the cache from disk, or crawls dalamud.dev if missing/outdated. Also checks the Dalamud GitHub releases in the background — if a new version is detected, the cache rebuilds automatically.
Cache versioning: the cache carries a version field. If the on-disk cache was built by an older version of this server, it is discarded and rebuilt on the next startup.
Lazy member loading: namespace and type indexes are cached upfront. Individual member details (properties, methods, etc.) are fetched on demand the first time get_type is called for a type, then persisted to cache.
Development
npm run dev # TypeScript watch mode
npm test # Run tests (Vitest)
npm run build # Compile to dist/
Tests cover search (including kind filter), searchMembers, findType, findNamespace, isEventRelated, and isValidCache — the core pure-function logic. The crawler itself (network I/O) is not unit-tested.
Project structure
src/
index.ts # MCP server, tool handlers
crawler.ts # Crawling, caching, types
search.ts # findType, findNamespace, search, searchMembers, isEventRelated
type-members.ts # groupMembersByKind, MEMBER_KIND_ORDER
__tests__/
crawler.test.ts
search.test.ts
cache/
dalamud-api.json # generated, not committed
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。