Convert your manuscript to markdown

Scriptorium works best on markdown-flavored manuscript text. This guide covers the common source formats. Pick the section that matches yours.

Already markdown (Quarto, plain `.md`, Pandoc Markdown)

No conversion needed. Skip to Install and start running skills.

Recommended: pandoc.

pandoc manuscript.docx -o manuscript.md --wrap=preserve --markdown-headings=atx

Alternative: mammoth (better for Word-styled documents).

mammoth manuscript.docx --output-format=markdown > manuscript.md

What you lose: tracked changes, comments, complex tables, embedded images-as-objects. Re-resolve manually if relevant.

pandoc manuscript.tex -o manuscript.md \
  --bibliography references.bib \
  --citeproc \
  --wrap=preserve

What you lose: custom macros; complex math environments survive but rarely render in markdown viewers; TikZ figures need separate export.

File → Download → Markdown (.md) — built-in since 2024. Or download as Word and use the .docx instructions above.

Use the LaTeX instructions above on the project’s main .tex.

Quality varies. Try:

pdftotext -layout manuscript.pdf manuscript.txt

Or marker / nougat for academic PDF OCR. Expect manual cleanup. Citations, figure refs, and table structure usually need re-resolution.

Validate that citations and figure references survived (a quick diff against the source helps).
Populate MANUSCRIPT_STATE.yaml with project.source_format: set to the original format (docx-via-pandoc, latex, gdocs-export, etc.) — this is a hint for skills that may apply format-specific parsing in v0.2+.
Run scriptorium validate <state-file> before running any skills.