cast-report

Cast report (landscape PDF + DOCX)

You are turning a cast CSV into a formatted, send-ready report for the screenplay in ${CLAUDE_PROJECT_DIR}. The CSV is a per-role list — Role, Description, Performer — typically a scene-scoped export from the writer's app (Fathom/Highland/etc.) or a derived doc. The output is a letter-landscape, 10pt, bold-header report as both PDF (for sending) and DOCX (editable), with each role's row kept whole across page breaks.

Preflight

1. Get the cast CSV path from the writer (or find the scene-scoped cast CSV they exported). It needs Role / Description / Performer columns (case-insensitive; extra columns ignored).
2. Note two optional sources the script uses if present: Claude Docs/casting_editorial.json (to backfill blank descriptions) and the project's Conversion Tools/md2docx.js (for the DOCX). The PDF needs Google Chrome (headless render so page-breaks / repeating headers / landscape are honored); if Chrome is absent it falls back to Conversion Tools/html2pdf.swift with a warning (that path can split long rows).

Run

3. python3 "${CLAUDE_PLUGIN_ROOT}/scripts/make_cast_report.py" --csv "<the CSV>" --title "<Project> — Cast …" --subtitle "<scope, e.g. Scenes 1–30>" with CLAUDE_PROJECT_DIR set to the project. Add:
   • --blank-performer for an outbound-to-casting sheet (empties the Performer column);
   • --no-fill to leave blank descriptions blank instead of pulling from casting_editorial.json;
   • --out-stem "<name>" to set the output filename (default: <csv> - Report beside the CSV).
4. The script writes <stem>.pdf, <stem>.docx, and the .html/.md sources. It never invents data — a blank field with no editorial fallback stays blank.

Report

State the output paths, the role count, and whether the PDF used Chrome or the Swift fallback (so the writer knows if rows could split). Do not commit — these are deliverables the writer reviews and sends; leave them for the writer to keep or discard.