ai-model-cloudbase_skill

Overview

This skill is a practical guide for calling AI models with CloudBase across web, Node.js, HTTP, and WeChat Mini Program platforms. It covers JS/Node SDK initialization, text generation (streaming and non-streaming), image generation on Node, and the HTTP/OpenAI-compatible gateway. The goal is to help developers pick the correct API shape and implement robust, platform-specific integrations quickly.

How this skill works

It explains SDK initialization differences (web vs Node vs wx.cloud), the shared AI API for JS/Node, and the distinct WeChat Mini Program API shape. It shows how to call generateText, streamText, and generateImage (Node only), and how to use the CloudBase HTTP gateway and OpenAI-compatible client. The guide highlights return formats, streaming iteration, callbacks for Mini Program, and token/usage reporting.

When to use it

• Building chat or assistant features that call CloudBase-hosted AI models from browser or server.
• Implementing streaming responses for long-form or interactive text generation.
• Generating images server-side using CloudBase Node SDK (image API is Node-only).
• Connecting third-party OpenAI-compatible clients to CloudBase via the HTTP gateway.
• Integrating AI in WeChat Mini Program where the API shape and callbacks differ.

Best practices

• Choose streaming for long responses to improve UX and reduce perceived latency.
• Wrap all AI calls in try/catch and surface friendly errors to users.
• Never embed secret keys in client-side code; use server or cloud functions for sensitive calls.
• Initialize CloudBase SDK or wx.cloud at app entry point to reuse clients.
• For Mini Program streaming, wrap parameters in data and check event.data === "[DONE]" to detect completion.

Example use cases

• Browser chat widget using @cloudbase/js-sdk with generateText and streamText for live replies.
• Server-side batch prompt processing and image generation using @cloudbase/node-sdk in cloud functions.
• Mobile app backend calling the CloudBase HTTP gateway via OpenAI-compatible client for unified API usage.
• WeChat Mini Program chat feature using wx.cloud.extend.AI.streamText with onText and onFinish callbacks.
• Edge function that proxies requests, keeps API keys secure, and returns streaming chunks to clients.