N°006 — THE HUMAN STORY

I build local-first agents and developer tools that make machine intelligence easier to inspect, trust and control.

Current position

我是唐万桂，一名把个人项目当作长期系统来构建的学生开发者。现在的重点是 AI Agent、开发者工具，以及人在复杂自动化中如何保持知情和控制。

I work across product definition, interaction, protocols, backend reliability and client implementation because trust is usually lost between those layers—not inside just one of them.

A shorter, factual viewOpen the résumé

Working principles

Local before remote

Keep useful behavior available when a provider, network or account is unavailable.

Evidence before confidence

Attach source, validation and uncertainty instead of making the interface sound more certain than the system.

Control before autonomy

Let software become proactive only after its permissions and consequences are visible.

See these principles in motionEnter Mahiru OS

Systems in progress

GitHub ↗Resume ↗Contact ↗