The Document Engineering textbook series (Vol 0– 5: Foundation, File Systems, Markdown, LaTeX, Citations, Diagrams), for Vietnamese learners aged 14 and up, ships its first edition (2026-04-25) with a placeholder tools/ directory and an explicit deferral of companion artefacts to a later session. The deferral is a sound scope decision: building companion artefacts in parallel with six volumes of full-draft authoring would have widened Phase 2 beyond solo-built capacity. The deferral is also a debt: when the artefacts are promoted to active, an architecture that the deferred decision pointed at must already exist, or the active phase reproduces the design work in haste. This paper supplies that architecture as a system specification with seven parts: an artefact taxonomy (templates, vault structures, style guides, exercise banks, sample documents, slide decks, validator scripts), a cross-volume coherence ruleset, a versioning and evolution discipline keyed to the tool half-life cycle, a storage and access model sized for a solo-built program, a concrete tools/ directory layout, a promotion-from-deferred protocol with entry conditions and exit criteria, and an audit rubric sized for one auditor to score one artefact in roughly twenty minutes. The synthesis draws on textbook-ancillarycontent research, open-textbook companion patterns, softwareengineering template-system architecture, the Microsoft and Google developer-documentation style-guide tradition, the Obsidian vault-design community resources, the DITA / structuredauthoring literature, and the Carpentries multi-curriculum companion infrastructure as an operational reference. The specification is engineered to preserve the existing deferred status (the architecture does not silently activate the node) and to make promotion-to-active a documented protocol rather than an adhoc decision. Three open questions are flagged for downstream research: template synchronisation across Obsidian Markdown, raw Markdown, and LaTeX representations; retirement when a tool substrate disappears; and whether a separate audit-loop track is required for companion-artefact updates that lag the book's audit cycle.
That Le (Tue,) studied this question.