dsh-user-mirror: Help AI Remember How You Think, Not Just What You Say

`dsh-user-mirror` is a DSH plugin designed to address the issue of models forgetting user principles and work styles across sessions. It analyzes model reasoning chains to proactively record users' judgment rationale (such as principles and red lines) and reuses this information in the system prompt within a token budget. It supports a 30-day half-life decay mechanism, profile display, and local storage, but does not currently support queries or Chinese synonym optimization.