Skip to content

ADR-v2-110 — Screenplay & Dialogue Auto-Format (Fountain)

Status: Accepted (2026-07-02) · Wave M Context links: markup (generic Markdown), [[adr-v2-111-semantic-line-breaks]], [[adr-011]]

Context

Wave M research (#6) — dictating "scene: interior coffee shop, day … Maya (character) quote where were you unquote" auto-emits correctly structured Fountain (scene headings, character cues, parentheticals, smart curly quotes, blank-line separation). Structured-Markup Dictation handles generic Markdown/headings; no excluded feature knows the screenplay/dialogue grammar (speaker turns, cues, transitions) or does smart-quote dialogue attribution. A new output-formatting domain. Anchor: Fountain plain-text screenwriting markup (fountain.io) — "indistinguishable from Final Draft when exported", Markdown-inspired and text-editor-native, so a formatter core is pure string transformation.

Decision

Add an opt-in Screenplay Auto-Format: [screenplay] enabled=false. Pure cores in screenplay/fountain.py: to_fountain(utterance) (detect scene heading → INT./EXT. LOCATION - DAY, "NAME (character) …" → an uppercased cue + dialogue, "transition: cut to" → CUT TO:, else an action line) and smart_quote_dialogue(text) (spoken "quote/unquote" and straight "…" → curly quotes). Pure regex/string transformation; no model. OFF by default.

Consequences

  • Hands-free screenplay authoring in a standard, portable plain-text format.
  • Pure transformation → fully testable.
  • Distinct from Structured-Markup (screenplay grammar vs generic Markdown).
  • Privacy (ADR-011): local text only.
  • Caveat: heuristic element detection (a full parser is a later tier); off by default.