可爱的、视觉化的 AI 构建者
要求 AI 将项目源代码导出或文档化，识别生成的假设，列出移动设备上仍然缺失的资产和应用身份。
问题
AI 构建者可以制作演示。他们需要准确的手机指示.
- AI 在浏览器停止工作，除非您告知它
- 响应式 Web 应用程序不是可安装的应用程序。AI 必须被告知考虑应用程序身份、原生权限、图标、启动屏幕、插件需求和设备行为。
- 原生发布工作仍然需要交互
- 苹果签名、安卓密钥库、商店记录、原生同步和生产构建工件通常发生在 AI 构建者之外。 AI 应该记录这些缺口而不是隐藏它们
- 更新规则需要明确
- 第一个二进制文件只是开始。 AI 应该定义哪些属于 Web 层更新、哪些需要新的原生构建，以及生产、预览和预览更新如何保持分离
Source requirements
首先告诉 AI 需要检查的源代码
只有 AI 认识到当前应用的实际源代码、构建输出、资产和本机假设时，它才能准备出有用的移动交接文件。
Bolt、Cursor 和基于仓库的工具
基于仓库的助手可以直接修改文件，但仍应返回可审查的移动交接文件，而不是假设本机发布工作已完成。
Base44、Next.js、Vite 和现有的 Web 应用
现有的应用也需要相同的检查：构建输出、路由、认证重定向、资产路径、本机插件需求和发布通道边界。
The Solution
Before Mobile Release
The AI cannot complete the store release alone. It can still prepare the repository and leave a precise checklist for the native release step.
- Inspect the Web App First
- Ask the AI to identify the framework, build output, routing model, environment variables, native plugin needs, and files that determine whether the app can live inside a WebView.
- Prepare Mobile Configuration
- Have the AI write the desired app ID, app name, webDir, icon and splash inputs, permissions, plugin needs, and missing native assumptions without pretending secrets exist.
- Plan Update Boundaries
- The AI should separate web-layer changes from native-store work, then document which future updates can move through release channels after the first app is installed.
- Create a Real Handoff
- The final output should state what the AI changed, what it could not do in the builder, and what a human or release service must handle next.
AI Instruction
给 AI 提供这个精确的移动发布说明
将此内容粘贴到 AI 构建器中，之后 web 应用程序工作。它告诉代理程序检查、文档化并为移动发布做好准备，而不假设本地工具访问。
AI 移动发布说明
You are preparing this existing web app to become an installable iOS and Android app.
Do not assume this builder can run local tooling. If this environment cannot create native projects or run local tools, create a clean handoff document instead.
Inspect the app and return:
1. The framework, package manager, production build process, and build output folder.
2. Whether the app can be exported or committed to a real Git repository.
3. The app name, bundle ID or package ID suggestion, app icon and splash-screen source, and any missing brand assets.
4. Auth, redirect, deep-link, storage, camera, push, payment, geolocation, file, or notification features that need native plugin planning.
5. Routing, environment variables, API URLs, and asset assumptions that could break inside a native WebView.
6. A proposed Capacitor configuration, including appId, appName, webDir, and required plugins, without inventing secrets.
7. A release-channel plan: production, staging, preview, and rollback expectations for web-layer updates.
8. A list of tasks that must happen outside the AI builder, such as Apple signing, Android keystore setup, native sync, store records, and signed builds.
9. A docs/mobile-release-handoff.md file with the exact checklist, open questions, and files that a developer or build service should handle next.
10. A short summary of what you changed, what you could not do in this environment, and the next human action.手动检查清单
AI 应该返回什么
目标不是一键自动化。目标是清晰、可审阅的发布手册，开发人员或发布服务可以继续。
1
检查当前应用
要求 AI 识别框架、构建输出、路由模型、资产、环境变量和本机敏感特性。
2
准备移动计划
要求它草拟应用标识、Capacitor 配置、插件需求、权限、图标和启动屏输入，以及缺失的本机假设。
3
编写发布手册
要求一个 docs/mobile-release-handoff.md 文件，包含更改的文件、开放的问题、未被 AI 构建器发明的机密以及外部 AI 构建器的任务。
4
检查发布边界
通过 web 层更新什么、需要原生构建什么以及等待签名或商店访问的工作需要哪些
用户信号
当 AI 不被要求神奇地发布移动应用时，结果最强大。它被要求检查 web 应用、准备配置并为发布步骤写好手动。
常见的 AI web-to-mobile feedback