html

Paged HTML reader

You are generating the paged HTML reader for the screenplay in ${CLAUDE_PROJECT_DIR} — the standalone web version (e.g. for embedding in Squarespace), an alternative to a sketchy PDF embed.

Inputs

1. Read ${CLAUDE_PROJECT_DIR}/Claude Docs/PROJECT_PROFILE.md §0 for source.text_mirror (the title-page + body text the converter takes; NOT the body-only .fountain, so the title page renders). If the mirror is stale, run /screenplay-prep:reconcile first (or refresh it) so the HTML reflects the current script.
2. The converter is ${CLAUDE_PLUGIN_ROOT}/scripts/script_to_html_paged.py. It is fully arg-driven (--input, --out, --page-height) and Highland-calibrated (parenthetical width 212 for Highland page parity — deliberately NOT the Fathom app's 137; don't "fix" it). [[ … ]] notes render as distinct .note boxes — these are intentional narrator notes the writer exports in the Highland PDF, kept visible, not hidden.

Run

3. Run python3 "${CLAUDE_PLUGIN_ROOT}/scripts/script_to_html_paged.py" --input "<text_mirror>" --out "<project>/Claude Docs/generated/script_paged.html". The generated/ tree is conventionally gitignored.
4. Report the output path and the page count the converter landed (it should match the canonical body-page count in §0 canonical.body_pages if one is set). If it diverges by more than a page, flag it — usually a pagination-calibration or notes-handling issue, not a content one.

Note

The project's git pre-commit hook may already regenerate this HTML on every screenplay commit; this skill is for an on-demand refresh or a first generation. Do not commit the generated file unless the writer asks — it's a derived artifact.