TDX MCP Server
An MCP server that wraps the TeamDynamix Web API, enabling AI assistants to search tickets, manage assets, query the knowledge base, and look up people in your TDX instance.
README
TDX MCP Server
An MCP (Model Context Protocol) server that wraps the TeamDynamix Web API, enabling AI assistants to search tickets, manage assets, query the knowledge base, and look up people in your TDX instance.
Quickstart
Prerequisites
- Bun v1.3+
- A TeamDynamix instance with Web API access
- Admin service account credentials (BEID + Web Services Key)
1. Install dependencies
bun install
2. Configure environment
cp .env.example .env
Edit .env with your TDX instance details:
TDX_BASE_URL=https://yourinstance.teamdynamix.com/TDWebApi/api
TDX_BEID=your-admin-beid-guid
TDX_WEB_SERVICES_KEY=your-web-services-key-guid
TDX_TICKETING_APP_ID=123
TDX_ASSET_APP_ID=456
TDX_KB_APP_ID=789
Finding your App IDs: If you don't know your application IDs, you can discover them after setup using the tdx_applications_list tool, or by calling GET /api/applications on your TDX instance directly.
3. Run the server
bun run src/index.ts
The server communicates over stdio using the MCP protocol. It's designed to be launched by an MCP client (like Claude Desktop or Claude Code), not run standalone.
4. Connect to an MCP client
Claude Desktop (~/.claude/claude_desktop_config.json):
{
"mcpServers": {
"tdx": {
"command": "bun",
"args": ["run", "/absolute/path/to/src/index.ts"],
"env": {
"TDX_BASE_URL": "https://yourinstance.teamdynamix.com/TDWebApi/api",
"TDX_BEID": "your-beid",
"TDX_WEB_SERVICES_KEY": "your-key",
"TDX_TICKETING_APP_ID": "123",
"TDX_ASSET_APP_ID": "456",
"TDX_KB_APP_ID": "789"
}
}
}
}
Claude Code (.claude/settings.json):
{
"mcpServers": {
"tdx": {
"command": "bun",
"args": ["run", "/absolute/path/to/src/index.ts"]
}
}
}
When using Claude Code, place your credentials in .env at the project root (it's gitignored). Bun loads .env automatically.
Running tests
# All tests (unit + QA)
bun run test
# Unit tests only
bun run test:unit
# QA/integration tests only
bun run test:qa
Unit and QA tests run in separate processes to avoid mock.module cross-contamination.
Project structure
src/
index.ts # Entry point — stdio transport, graceful shutdown
server.ts # MCP server — tool registration
config.ts # Env var loading with Zod validation
auth/
client.ts # TDX auth — JWT caching, proactive refresh
schemas.ts # Auth Zod schemas
api/
client.ts # Fetch wrapper — auth headers, retry on 429
rate-limiter.ts # Client-side sliding window rate limiter
tools/
tickets.ts # Ticket tool handlers
assets.ts # Asset tool handlers
kb.ts # Knowledge base tool handlers
people.ts # People tool handlers
applications.ts # Applications tool handler
utils.ts # safeToolCall error wrapper
schemas/ # Zod input schemas for each tool
tests/
config.test.ts # Config loading tests
server.test.ts # Server creation tests
auth/ # Auth client tests
api/ # API client + rate limiter + error scenario tests
tools/ # Tool handler tests
qa/ # End-to-end MCP server tests
fixtures/ # Realistic TDX API response fixtures
Authentication
The server uses TDX admin authentication (POST /api/auth/loginadmin) with BEID and Web Services Key. The JWT is cached in memory and refreshed automatically 5 minutes before expiry (tokens last 24 hours).
Rate limiting
TDX enforces per-endpoint rate limits with no response headers. The server tracks limits client-side using a sliding window algorithm. Limits vary by endpoint:
| Endpoint | Limit |
|---|---|
| Ticket search | 30 / 60s |
| Ticket create | 120 / 60s |
| People CRUD | 45 / 60s |
| People lookup | 75 / 10s |
| Most others | 60 / 60s |
On 429 responses, the server retries up to 3 times with backoff.
Tech stack
- Runtime: Bun
- Language: TypeScript (strict mode)
- MCP SDK: @modelcontextprotocol/server
- Validation: Zod
- Linting/Formatting: Biome
- Testing: bun:test
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。