ai-model-wechat

What it does

Instructs an AI agent to call WeChat Mini Program AI models through wx.cloud.extend.AI, choosing a provider (hunyuan-exp or cloudbase) and a concrete model ID placed in the data field. It includes two-step preflight: ① Billing Eligibility (determine whether to use hunyuan-exp or cloudbase) and ② Group Readiness (enable the target model in DescribeAIModels / DescribeManagedAIModelList / UpdateAIModel). It also covers how to query and enable models, and cautions against using non-supported provider names or model IDs in createModel.

How it works

• Before code: perform Mandatory Two-Step Preflight in order: ① Billing Eligibility by querying env and enrollment (growth plan vs token credits) and selecting provider: "hunyuan-exp" for Growth Plan or "cloudbase" for Token Credits; the chosen path determines default or selectable models. ② Group Readiness: query DescribeAIModels for the EnvId; if the target model is not present, query DescribeManagedAIModelList to find the correct Model string, then enable via UpdateAIModel with Status: 1 and a full replacement Models[] including the target model. Only after both steps pass, instruct to write wx.cloud.extend.AI.createModel("<GroupName>") and ensure the data field contains a valid model.
• It instructs exact provider/model separation: provider (GroupName) goes into createModel("provider"), while the concrete model id goes into data: { model: "..." } within streamText/generateText calls.
• It defines the three legal provider shapes and shows example patterns for both hunyuan-exp and cloudbase usage with the proper data.model field.

When to use it

Use this skill for calling AI models in WeChat Mini Program using wx.cloud.extend.AI. Use when you need to generate or stream text with onText / onEvent / onFinish callbacks, and when selecting models from either the Growth Plan or the main managed cloudbase group. Do not use for browser/Web, Node.js backend, or image generation in Mini Program contexts.

What it can touch

• It references and uses CloudBase API calls through the tcb service: DescribeAIModels, DescribeManagedAIModelList, UpdateAIModel, DescribeEnvPostpayPackage, CreateAIModel. It requires EnvId, GroupName, Model, and data structures like { model: "...", messages: [...] } for streamText/generateText. Commands and file names are quoted exactly as in the skill, e.g. wx.cloud.extend.AI.createModel, DescribeAIModels, UpdateAIModel, DescribeManagedAIModelList, DescribeAIModels, DescribeEnvPostpayPackage.

Caveats

• The provider argument to wx.cloud.extend.AI.createModel(provider) must be one of three legal shapes: "hunyuan-exp", "cloudbase", or "custom-<name>". Do not pass model IDs in the provider slot. The model ID goes into the data.wrapper field as { model: "..." }.
• The two-step preflight must be completed before any business code is written.
• Enabling models requires full replacement of the Models[] array in UpdateAIModel (not incremental).
• The skill distinguishes between Growth Plan vs Token Credits, and warns to enable the exact target model before invocation; otherwise ModelNotEnabled errors may occur.