AiToEarn details multi-platform publishing adapter architecture
The AiToEarn team has documented an open-source MIT-licensed system that combines platform adapters with a registry to unify authorization, publishing, status checks, engagement, and analytics. Its architecture declares authorization and completion methods explicitly, while BullMQ workers, Redis persistence, and Redlock-based idempotency support asynchronous delivery without duplicate posts. The project also separates the publishing service from AI orchestration and packages MCP tooling as a NestJS module.
The source provides a concrete, source-inspectable reference for handling inconsistent platform authorization, asynchronous publication, and duplicate prevention in multi-platform publishing workflows.
当一个平台不给你 OAuth 时:多平台内容分发的适配器设计
做多平台内容分发,最麻烦的从来不是"发"这个动作,而是十几个平台各自为政:授权方式不同、内容格式不同、发布结果什么时候算"成功"也不同。
我们团队在做这件事时,把这些差异收敛成了一套「适配器 + 注册表」结构。这篇文章拆解它的设计,重点讲三个很少有人提、但真正决定这套系统能不能跑的点:
- 平台不给你 OAuth 怎么办
- 发布结果怎么算"成功"
- 异步流水线怎么保证不重复发
代码是开源的(MIT),下文所有结论都能通过读源码验证。
一、接口层:五类能力契约
先说最核心的抽象。platforms.interface.ts 里定义了五类能力:
#AiToEarn#BullMQ#Redis#Redlock#NestJS#Model Context Protocol