Screenplay Scene
Write a properly formatted screenplay scene with a clear want, escalating conflict, subtext-driven dialogue, and a turn at the end.
Write a screenplay scene in standard US format. Story context: [WHAT THE AUDIENCE ALREADY KNOWS BY THIS POINT] Characters in the scene: [NAME, AGE, AND THE ONE THING THEY WANT RIGHT NOW] Location and time: [WHERE, WHEN, INT or EXT] What each character wants from the other in THIS scene: [THE COMPETING WANTS] What must be true when the scene ends: [THE PLOT REQUIREMENT] Genre and tone: [e.g. "grounded drama" / "workplace comedy" / "slow-burn thriller"] Length: [PAGES, roughly one page per minute of screen time] Craft rules: - Enter the scene as late as possible and leave as early as possible. Cut the greeting and the goodbye. - Every scene is a negotiation. Both characters want something incompatible, and the balance of power shifts at least once. - Characters say what serves their want, not what they feel. Put the feeling in the choice of subject, the deflection, the thing they refuse to answer. - No character explains the plot to another character who already knows it. - Action lines: present tense, only what the camera sees and the microphone hears. No inner thoughts, no backstory, no camera directions. - Give each character a distinct rhythm: sentence length, vocabulary, whether they interrupt, whether they finish sentences. - End on a turn: a decision, a revelation, or a shift in the relationship. Not a summary. - Use parentheticals rarely, only when the line would otherwise be read wrong. Format exactly: INT. LOCATION - DAY Action line in present tense. CHARACTER NAME (parenthetical, only if essential) Dialogue. Return the scene only. Then, separately, one line naming the shift in power and where it happens.
How to use
The competing-wants field is what separates a scene from a conversation, so fill it in even when the answer feels obvious. Models default to on-the-nose dialogue where characters announce their feelings, which is why the subtext rule is stated as a mechanism ('put the feeling in what they refuse to answer') rather than as the word 'subtext'. The power-shift line at the end is a self-check: if the model cannot name where it happens, the scene is flat and needs a rewrite, not an edit.
More writing prompts
Rewrite the text below in plain language for [AUDIENCE, e.g. "patients with no medical background" / "customers reading a policy update" / "new hires on day one"]. Target reading level: [e.g. "8th grade" / "as simple as the content allows"]. How to rewrite: - Lead with what the reader must DO or what happens to them. Background comes af
Plain Language Rewrite
Rewrite dense policy, legal, medical, or technical text at a target reading level without losing a single obligation or number.
Write a comparison article for the query: [e.g. "Notion vs Obsidian" / "best Mailchimp alternatives"]. Options to compare: [LIST 2-8 OPTIONS] Who is searching this: [READER + WHAT THEY ARE TRYING TO DECIDE] My relationship to these products: [e.g. "I make one of them" / "I use two of them daily" / "none, I am researching"] Facts I can su
Comparison Article
Write an honest X vs Y or best alternatives article: real differentiators, a decision table, and a clear pick for each type of reader.
Build an FAQ page for [PRODUCT/SERVICE/TOPIC]. Real questions people ask me: """ [PASTE SUPPORT TICKETS, SALES CALL NOTES, SEARCH QUERIES, OR COMMENTS. RAW IS FINE.] """ Facts I can state as true: [PRICING, POLICIES, LIMITS, TIMELINES] Do this: 1. **Cluster** the raw questions into distinct questions people actually ask. Merge duplicat
FAQ Page From Real Questions
Turn support tickets and sales objections into an FAQ page that answers in the first sentence and is structured for AI search engines.