An AI agent traces China Telecom’s Home Assistant authorization flow
The author used a local AI agent to analyze China Telecom’s authorization protocol and develop telecom-phone-device-id-tools for a Home Assistant integration that initially failed with resultCode 3006, “设备未信任.” The article documents the local-memory authorization workflow, slider-verification alignment, entity-registration fixes, and testing with two phone accounts.
The account demonstrates a practical use of a local AI agent for protocol analysis and Home Assistant integration work while keeping sensitive credentials out of third-party hosting.
把电费账单接进 Home Assistant 之后,我自然的下一步,就是把全家手机的话费余额与通用流量也放上家庭概览大屏。 查阅社区插件,填入手机号与电信服务密码,点击提交。等待我的不是清爽的余额图表,而是一行冰冷的报错:resultCode 3006,设备未信任。 查阅社区讨论,有人建议使用网关免密,有人甚至推荐把手机号、服务密码和短信验证码提交给第三方的未知反代托管网站。把掌握电信业务核心控制权的服务密码交由陌生服务器中转?这是任何重视家庭隐私与底层安全的人都绝不能触碰的红线。 于是,今天我让本机 AI Agent 彻底拆解了中国电信官方登录协议链:从抓包验证家庭宽带免密为何失效,到手搓一个仅在本地内存运行、零外部云端依赖的局域网授权工具;从离屏 Canvas 像素扫描攻克滑块拼图的几何漂移陷阱,到重构 Home Assistant 集成生命周期解决首刷崩溃与并发竞态。 这是一次历时整整一天的软硬件与协议攻防实录。