> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
**Goal:** Export every authoring-content block in a book as readable Word-compatible HTML.
**Architecture:** The backend retains the complete ordered chapter tree and includes every node's body before its children. A focused frontend normalizer replaces editor-only custom nodes with static HTML, loads practice questions through the existing API, then passes the result to the existing image-embedding DOCX exporter.
-[x] Change the export tree traversal so it appends a node's content in a chapter-id container immediately after its heading, before recursing into children.
-[x] Return a base export name without a misleading `.pdf` extension.
-[x] Write failing tests proving gallery images/captions, fetched expansion text, links, tooltips, media links, and a fetched practice question become static HTML.
-[x] Implement a DOM-based normalizer that converts each custom node while preserving ordinary body HTML and loads dynamic extension/practice payloads from injected request functions.
-[x] Run the focused Vitest test and confirm it passes.