documents — written material in and out

Home branch in Memora: root/skill/documents.

Choosing a part

Part	What it covers
parts/md-document/	Long-form markdown (specs, RFCs, reports, plans, explainers) → a single-file, lightly interactive HTML document with sticky TOC and scrollspy
parts/markdown-html-orchestrator/	The general markdown → single-file HTML converter, for any markdown in a project
parts/md-review/	A PR writeup or code review with diff blocks and [!BLOCKER]/[!MAJOR]/[!MINOR]/[!NIT] callouts → a readable review document
parts/read-docs/	Reading .doc/.docx quickly via python-docx
parts/diagram-generator/	Diagrams from natural language, notes, code, schemas, tables or existing source

PDF, Word-creation and spreadsheet engines are not bundled. The usable ones are source-available under terms that forbid repackaging, so this kit ships neither (see PROVENANCE.md). When the deliverable must be an .xlsx, a filled PDF form or a Word file with tracked changes, install that engine from its own source first and record the install there — do not hand-edit a zip of XML.

Rules

• HTML output is one file, opened locally. No build step and no CDN dependency for something the user will send to someone; check it in a browser before calling it done.
• Office files are edited by a tool, not by hand. A re-zipped XML rewrite is how documents get corrupted; if no engine is installed, say so and install the right one instead of imitating it.
• Prose quality is not optional: long documents go through antislop before delivery.
• Ask when it matters: the output format when the request could mean HTML or a Word file, a document that will be signed or submitted formally, and any file destined for someone outside the workspace.
• What the document decided becomes a memory node under root/skill/documents; the file itself belongs with its project.

Related

decks (slides), antislop (prose), documentation-writer (formal reports), design (typography and layout of the rendered page).