APIs and datasets: pull config, see your data clearly

This page explains how to have an Agent align your local API config, explore platform datasets, and confirm fields and operations — all in Vibe Coding. Don't assemble commands by hand; describe the business task and let the Agent inspect, sync, and explain the results.

Tasks to hand to the Agent

• Keep the current project's data API config in sync with the platform.
• Find the dataset behind a business object.
• Confirm a dataset's fields, primary key, query operations, and input parameters.
• Map out table relations across datasets as preparation for SQL, BFF, or page work.
• Make controlled changes to a few fields' display info or dataset metadata.

Prerequisites

• The target project is open in your Vibe Coding tool.
• rabetbase is installed and you're logged in.
• You know the business object for this session — for example, orders, customers, contracts, or tickets.

Common prompts

Sync local API config

请帮我检查当前项目的数据接口配置是否和平台一致。
如果不一致，请同步本地 API 代码。
完成后告诉我：新增、删除或变化的模型有哪些，前端代码可能需要关注哪些字段。

The AI may call: rabetbase api pull, rabetbase api list.

Find the dataset for a business object

请帮我找一下和「客户」相关的数据集。
请说明每个数据集的用途、关键字段、主键、常用查询操作。
如果有多个候选，请推荐最适合用于「客户列表页」的那个，并说明理由。

The AI may call: rabetbase dataset list, rabetbase dataset detail.

Confirm how to call a dataset

请帮我查看「客户」数据集的数据结构。
重点确认客户列表页的关键词搜索、客户等级筛选、负责人筛选、分页查询和客户详情查询分别需要哪些参数。
返回字段请重点说明客户名称、联系人、手机号、客户等级、跟进状态、负责人、最近跟进时间。
请按前端开发可直接理解的方式说明，不要只贴原始字段。

The AI may call: rabetbase dataset detail, rabetbase dataset operations, rabetbase api pull.

Map table relations

我要基于「客户列表页」做客户、联系人、跟进记录的组合展示。
请先梳理客户、联系人、跟进记录之间的关系，包括主表、关联表、关联字段和容易用错的字段。
最后给出适合后续写 SQL 或 BFF 的数据准备说明。

The AI may call: rabetbase dataset links, rabetbase dataset detail.

Make a controlled field display change

请把「客户」数据集里的「客户级别」字段展示成「客户等级」。
执行前先预览影响范围，确认只修改这个字段的展示信息。
执行后请告诉我修改前后对比，以及是否还需要同步页面或调用代码。

The AI may call: rabetbase dataset detail, rabetbase dataset field-update, rabetbase api pull.

What to add to your prompt

• Business objects: customers, orders, contracts, tickets, and so on.
• Target feature: a list page, detail page, query API, or report.
• Field semantics: which fields must be shown, and which must not be misused.
• Change constraints: read-only checks, preview first, don't overwrite existing files, etc.

Completion checks

• The Agent can name the dataset it used and justify the choice.
• Fields, operations, and relations are explained in business terms.
• Local API config is synced, and the frontend can keep using the generated models or client.
• Any write was previewed before it was executed.

Next step: to turn a query into a platform resource, see "Custom SQL: from writing to execution". To wrap server-side logic, see "BFF scripts: local and remote in sync".