Introduction
The TeamDynamix MCP Server connects AI assistants like Claude to your TeamDynamix instance via the Model Context Protocol.
What it does
Instead of switching between Claude and the TDX web interface, you can ask questions and take actions directly in your AI conversation:
- “Search for open tickets about VPN issues”
- “Look up the KB article on password resets”
- “Find all assets assigned to the biology department”
- “Create a ticket for a new laptop request”
Key features
Lazy loading
Only 2 tools load at startup. Domain tools load on demand — keeping the AI’s tool context small and fast.
4 domains
Tickets, Knowledge Base, People, and Assets — with more coming.
Auto-discovery
App IDs for ticketing, assets, and KB are discovered automatically from your TDX instance.
Multiple transports
Run via stdio for Claude Desktop, or HTTP for remote deployments.
How it works
The server uses a decision-tree navigation pattern:
- On startup, only
tdx_navigateandtdx_statusare registered - When the AI needs ticket data, it calls
tdx_navigate("tickets") - The tickets domain tools (
tdx_tickets_search,tdx_tickets_get, etc.) load dynamically - The AI uses those tools to fulfill your request
This pattern keeps the initial tool footprint small — important for LLMs that have limited tool context windows.
Next steps
- Installation — get the server running
- Configuration — set up your TDX credentials