user-lavishifylisted

# user-lavishify Take the artifact-shaped content just delivered in this conversation and turn it into a **Lavish Editor** review surface: a self-contained HTML page opened in the operator's own browser, where they annotate elements / text ranges / Mermaid nodes and the feedback returns here structured. This is the **inverse** of building in Lavish from the start — normal delivery is plain chat; this escalates a specific output to Lavish only when spatial annotation beats typing. `$ARGUMENTS` If the argument is non-empty, render THAT (a named plan file, a section, "the comparison you just showed"). If empty, render the **last substantial assistant output** — a plan, comparison, table, diagram, report, or a list of proposed changes. If the last output was short or conversational, ask what to render instead of guessing. ## When to use vs. not - **Use** when the just-delivered content is dense/structured and the operator would otherwise have to quote many things to fix — multi-step plans, option comparisons, diagrams, big tables, a long list of proposed edits, or a UI/design mock. - **Don't** for short or conversational replies — chat is faster and zero context-switch (thinking-as-you-type is the default). Lavish costs a browser context-switch + artifact-authoring effort; only spend it when element-level annotation genuinely pays. ## Workflow 1. **Pick the content.** The last substantial output (or `$ARGUMENTS`). Keep the substance identical — this is a re-presentation, n