A script doctor should name the disease
“Raise the stakes.” “Make the character more active.” “The second act feels slow.” These notes may point toward a problem, but they do not tell a writer what mechanism failed, where it failed, or what revision would repair it.
Laper's official AI Script Doctor mode is built around a stricter sequence:
- diagnose the root problem;
- locate the lesion in an act, beat, sequence, scene, character arc, or line;
- give an executable prescription;
- protect the writer's voice and intended story;
- prioritize load-bearing changes before cosmetic notes.
It is an optional advisor skill, not a background process and not an autonomous author.
Seven lenses, chosen by the problem
The skill carries seven established dramaturgy systems at once:
| Lens | What it is used to inspect |
|---|---|
| Aristotle | Action, reversal, recognition, causality, and unity of plot |
| Syd Field | Three-act load bearing points and the midpoint |
| Robert McKee | Value turns, expectation gaps, scene and sequence design |
| Save the Cat | A higher-resolution fifteen-beat diagnostic ruler |
| Hero's Journey | Resistance, ordeal, transformation, and return |
| John Truby | Want, need, moral weakness, opponent, self-revelation |
| David Mamet | Dramatic action, tactics, subtext, and dialogue that pursues a goal |
These frameworks are not treated as seven checklists that every screenplay must satisfy. The mode is instructed to choose the blade that explains the lesion. A missing midpoint and a line of on-the-nose dialogue are different diseases.
For writers studying the underlying craft, our three-act structure guide and Save the Cat beat sheet explain two of those lenses in depth.
What an executable diagnosis looks like
A useful note has three layers.
Root cause: “The second act sags because the protagonist has no escalating want; scenes deliver information but do not force riskier actions.”
Location: “The lesion begins after the midpoint. The apparent victory changes no value and the opponent does not adapt.”
Prescription: “Turn the midpoint into a public false victory, let it expose the protagonist's hidden compromise, and make the next three scenes remove one safe option each. Plant the compromise in the Act One relationship scene so the reversal is surprising but inevitable.”
That prescription tells the writer what to change, where to change it, and why the mechanism works. The skill explicitly rejects empty encouragement and large piles of equally weighted notes.
It reads through tools, not intuition theater
The doctor can use Laper's structured screenplay tools to map an outline, inspect a specific scene, locate a stable node, read a bounded range, or review current cursor context. On a long draft, it should read in passes: outline first, then the scenes that carry the central turns, then dialogue or line context where the diagnosis requires it.
The full read mode has a real boundary of 800 nodes. If a screenplay exceeds it, the tool reports truncation. The doctor should not claim to have read unseen pages.
This context discipline is described fully in How Laper Gives AI Structured Screenplay Context.
Revision remains human-led
The doctor may write an alternate beat, a revised exchange, or a new scene turn when an example makes the prescription concrete. But the operating instruction is “doctor, not author.” It must strengthen the controlling idea of the writer's story rather than replace the subject, genre, or taste with its own.
If intent is unclear, asking a question is better than performing a confident misdiagnosis. If the problem is structural, fixing commas first is avoidance. If the problem is a single unclear tactic, proposing a new third act is overreach.
The writer decides whether to apply a prescription. Approved screenplay edits return through the active editor and Loro document like other explicit AI tool operations; advice does not silently become canon.
When to use Script Doctor
The mode is most valuable for:
- a passive protagonist whose choices do not drive consequence;
- a sagging middle or missing midpoint;
- an unearned climax or missing setup/payoff chain;
- flat stakes that never become personal;
- scenes that end on the same value with which they began;
- dialogue that explains feelings instead of pursuing an objective;
- a theme stated in speeches but not tested by action.
It is less useful when the writer only needs spelling, page layout, or a factual lookup. Use the smallest tool that fits the problem.
Explore the full AI screenwriting software model or start from the screenplay editor when the draft itself needs a home first.