knowing-what-you-know-now

評估已運行系統是否值得重寫、重構或繼續保留並提供建議。

在一个已经运行且按预期工作的系统上，给出保留、重构或重写的建议——目的不是立刻改造，而是基于现在已知信息做有根据的判断。用在你想在继续开发前评估是否值得把系统重做时，会先确认评估目标（PR、目录或模块），不清楚会主动问清。强调要“赚得起重写”，避免无谓的重构浪费时间和成本。

# Knowing What You Know Now

Run this on something that is **built and working as intended**. Evaluate whatever
target you're pointed at (a PR, a directory, a module, a diff); if none is given,
infer it from the current conversation, and if it's still unclear, ask which build
to evaluate — don't evaluate air.

The job is a judgment call, not a rewrite: **earn the rebuild**. Most of the time
the honest answer is "keep it" or "one targeted refactor" — a full rebuild has to
justify itself. The guiding move is the whiteboard question: *knowing everything
you know now, how would you build it?* The reason an elegant version can exist is
that **the learning is the collateral, not the code** — you n