Documentation

Export

Export copies the piece’s most refined Markdown into a standalone file you can edit, commit, or publish elsewhere. Pysar never posts for you.

By default that file lands in the project root. You can choose a different project-relative directory at init (pysar init --export-dir published) or by editing export_dir in .pysar/project. The MCP tool export_piece_to_root uses that setting; an optional export_dir argument overrides it for one call only.

When it runs

/ps exports automatically after the final stage it runs (usually humanize).

You can also export via the MCP tool export_piece_to_root (skills invoke this; you rarely call it by hand). Skills omit export_dir unless you asked for a one-off path — the tool result reports where the file landed.

What gets exported

Export picks the newest revision that exists for the piece, in this priority:

  1. humanize.md
  2. seo.md (if humanize has not run)
  3. sharpen.md
  4. staff-edit.md
  5. draft.md

A draft is required. Earlier stage files are not deleted or rewritten by export. Re-running export overwrites the previous file for that piece at the resolved destination.

Export currently copies that revision as-is. Research citation markers like [^shortname] may still appear in the exported file until mechanical resolve-at- export ships. Do not treat /ps --seo as the required cleanup for those markers — SEO packaging is a separate opt-in stage.

Piece IDs and filenames

  • Working files: .pysar/pieces/<piece-id>/
  • Exported file: <export_dir>/<piece-id>.md (or <project-root>/<piece-id>.md when export_dir is unset; same basename as the piece directory; the MCP tool response reports the path and word count)

<piece-id> is a machine-made Latin slug from the piece title. Non-Latin titles (for example Ukrainian, Japanese, Arabic) are transliterated so you do not get a bare template-* id. Authors do not invent the Latin slug by hand.

Next

If export fails or picks the wrong revision, see Troubleshooting.