documentlisted

# Deliverable documents (markdown → pandoc → docx/pdf) One honest toolchain: author clean markdown, convert with pandoc. No fake Office XML by hand. ## Procedure 1. Author the content as a standalone markdown file (scratch or requested path) with a pandoc metadata block: ```yaml --- title: "…" author: "…" date: "YYYY-MM-DD" --- ``` Structure: title → 3-6 line executive summary → body sections → appendix for detail. Tables for enumerable facts; prose for reasoning. 2. Check the converter: `pandoc --version`. - **.docx**: `pandoc in.md -o out.docx` (add `--reference-doc=<template>.docx` if the user has a house template). - **.pdf**: needs a PDF engine - try `pandoc in.md -o out.pdf --pdf-engine=typst` first, then `weasyprint`/`wkhtmltopdf` via HTML, then LaTeX if installed. 3. **Verify the artifact**: file exists, nonzero size; for pdf run `pdfinfo` (or open page count); mention anything that didn't convert (e.g. raw HTML blocks). 4. If pandoc is absent: deliver the clean markdown, say exactly what to install (`pandoc` + `typst` are enough on Arch), and offer to finish after. ## Guardrails - Content comes from real sources in the conversation/repo - for run/pipeline reports prefer `ritual report` output as the base when `.ritual/` exists. - Anything secret-looking in content gets redacted before it enters a shareable file. - Never claim a conversion succeeded without checking the output file.