Cloud AI User Guide

These guides cover talking to a cloud language model directly with TTMSMCPCloudAI. Start with Getting started if you have not sent a prompt yet.

Guides

Guide	Scope	API starting points
Chat and prompting	Service selection and API keys, the system/assistant/user roles, per-service models, temperature and `UseTemperature`, token limits, reasoning effort, web search, Claude prompt caching with `CacheControl`, local Ollama and llama.cpp servers, OpenRouter, and reading the response and usage.	`TTMSMCPCloudAI`, `TTMSMCPCloudAISettings`, `TTMSMCPCloudAIResponse`
Function calling	Declaring tools, typed parameters, enum and format constraints, nested objects and arrays of objects, reusable tool sets, parallel tool execution, and the tool execution handler.	`TTMSMCPCloudAITool`, `TTMSMCPCloudAIParameter`, `TTMSMCPCloudAIToolSet`
Files, assistants, speech, and transcription	PDF, Office, image, text and URL context; uploading and deleting stored files; the assistants and threads API for retrieval; text-to-speech; and audio transcription and translation.	`TTMSMCPCloudAIFile`, `TTMSMCPCloudAIAssistant`, `TTMSMCPCloudAI`

Cloud AI Tool Sets ships ready-made TTMSMCPCloudAIToolSet descendants — logging, file system access, dataset queries, and email — that plug straight into the AI property.
Cloud Image AI covers image generation against the same kind of cloud services.
MCP Client reaches a model through an MCP server, so the server's tools, resources, and prompts come along with it. Use TTMSMCPCloudAI when you want the model and nothing else.