Components and Libraries
TMS AI Studio splits into three areas: the components that build an MCP server and describe what it exposes, the client and transport components that connect to a server or an LLM service, and the cloud AI components that call a provider directly. Pick the area that matches the side of the conversation you are writing.
Building an MCP server
Expose your application to natural-language agents. Start with the server component, then describe what it offers through tools, resources, and prompts — or let RTTI attributes generate the whole surface for you.
| Component | Description | Start |
|---|---|---|
| MCP Server | JSON-RPC 2.0 MCP server with sessions, sampling, elicitation, long-running tasks, pagination, and change notifications. | Get started · TTMSMCPServer |
| Tools | Callable operations an agent can invoke, with typed input properties, structured output, and a fluent builder. | Get started · TTMSMCPTool |
| Resources | Readable content exposed by URI, including multi-segment URI templates and subscription-based update notifications. | Get started · TTMSMCPResource |
| Prompts | Reusable, argument-driven prompt templates a client can list and retrieve. | Get started · TTMSMCPPrompt |
| Attributes | RTTI attributes that turn ordinary classes and methods into an MCP tool, resource, and prompt surface with no manual registration. | Get started · TTMSMCPCustomAttribute |
Connecting to servers and services
The client side of the protocol: connect to an LLM service, spawn or reach MCP server processes, and choose how bytes move between the two.
| Component | Description | Start |
|---|---|---|
| MCP Client | Connects an application to an LLM service and to one or more MCP servers, with tool-call routing and a settings dialog. | Get started · TTMSMCPClient |
| MCP Transports | STDIO, Named Pipe, SSE, and Streamable HTTP transports for both server and client, plus a standalone HTTP server. | Get started · TTMSMCPTransport |
Calling cloud AI directly
Use these when the application itself is the AI consumer — no MCP server involved. They wrap provider APIs for chat, function calling, files, assistants, and image generation.
| Component | Description | Start |
|---|---|---|
| Cloud AI | Direct access to cloud AI providers for chat, function calling, file submission, assistants, speech, and transcription. | Get started · TTMSMCPCloudAI |
| Cloud AI Tool Sets | Ready-made tool sets that give a cloud AI model access to the file system, a dataset, e-mail, and logging. | Get started · TTMSMCPCloudAIToolSet |
| Cloud Image AI | Image generation across OpenAI, Gemini, Black Forest Labs, Stability, and Reve through one API. | Get started · TTMSMCPCloudImageAI |