# Short Story — Plot: Plot Skeletons
We spent the last three articles making people. On Monday, in ["AI as Character Generator,"](/harrys-desk/short-story-character-ai-as-character-generator) we learned to find the contradiction that makes a character worth following. On Wednesday, in ["Backstory Engines,"](/harrys-desk/short-story-character-backstory-engines) we built the buried history that gives every present action weight. On Friday, in ["The Character Interview,"](/harrys-desk/short-story-character-the-character-interview) we discovered how our characters sound when they are pushed. If you have been doing the homework, you now have a person who wants something, fears something, and lies to themselves about both.
This week we leave the character shop and enter the architecture of plot. Not plot as a rigid formula — the kind of beat sheet that promises a bestseller if you only fill in the blanks — but plot as the shape that lets desire meet resistance. A plot skeleton is the minimum structure a story needs before it can walk. It is not the story itself. It is the scaffold that holds the story while you build it.
Today we focus on the short story skeleton specifically: the inciting incident, the rising action, the reversal, and the resolution. We will use AI to generate options at each joint, and we will use our own judgment to decide which options belong to the character we have already made.
The Difference Between Plot and Story
It is possible to have a perfect plot and no story at all. A plot is a sequence of events. A story is what those events mean to someone. The plot of *The Great Gatsby* could be summarized as a man throws parties to win back an old lover and gets shot. That is not why the novel endures. It endures because the events rearrange Nick Carraway's understanding of longing, class, and illusion.
AI is excellent at generating plots. Given a prompt, it will produce events: a letter arrives, a secret is revealed, a bridge collapses, a gun goes off. The danger is that the events will feel like events happening to no one in particular. The skeleton needs a spine, and the spine is the character's want.
Before you ask the model for a plot, you must be able to state the character's want in a single sentence that includes risk. Not "Clara wants to repair the opera-house clock." That is a task. Try: "Clara wants to finish the opera-house clock before the town council votes to demolish the building, because completing the clock is the only way she can prove that something beautiful is worth saving." Now the want has stakes, time pressure, and a value. Plot can attach to that.
The Four Bones of the Short Story Skeleton
For a short story, I work with four bones. Each one is a moment where the character's situation changes in a way that tests the want.
Bone one: the inciting incident. Something arrives that disrupts the character's ordinary equilibrium. It is not yet the main conflict, but it makes the main conflict possible. The inciting incident should be specific, concrete, and impossible to ignore.
Bone two: the rising action. The character responds to the inciting incident, and each response raises the stakes or deepens the difficulty. In a short story, rising action is usually compressed into a few scenes, sometimes one sustained scene. Every action should make the final choice more costly.
Bone three: the reversal. The moment when the character's plan collides with reality. The reversal can be an external defeat, an internal revelation, or both. It is the point where the character's lie is exposed and a new choice becomes necessary.
Bone four: the resolution. Not necessarily a happy ending, but a decisive ending. The character has changed or refused to change. The want has been achieved, redefined, or abandoned in a way that reveals who the character has become.
These four bones are enough to hold a short story. If you try to add more, you may be writing a novella. If you have fewer, you may be writing a vignette. Both are valid forms, but this series is about stories that move.
Prompting for the Inciting Incident
The inciting incident is the doorway into the story. It must be plausible in the world you have built and disruptive enough to force action. I usually ask the model for several options, each of which connects to a different part of the character's backstory.
Here is a prompt shaped around Clara Wren:
> Clara Wren is a clock restorer in a dying railway town. Her private lie is that objects matter more than people because objects do not abandon you. Suggest three inciting incidents for a short story, each about 40 words. One should involve the opera-house clock, one a person from her past, and one a threatened object in her shop. Each should make it impossible for her to continue as before.
The model might return:
- •The town council announces the opera house will be demolished in thirty days unless a private buyer commits to restoration by dawn tomorrow.
- •Clara's estranged daughter appears at the shop with a suitcase and a clock she wants repaired, the same model Clara's mother once owned.
- •A water pipe bursts over Clara's workbench, destroying a clock she promised to finish for the funeral of a neighbor's child.
Each option is a doorway. The first tests her belief that saving objects can redeem a town. The second confronts her with the person she most wants to treat as an object. The third gives her a grief she cannot avoid by retreating into work. None of them are random accidents. Each one is chosen to press on the lie we identified during backstory work.
Generating the Rising Action
Once the inciting incident is chosen, the rising action must make the character's chosen path increasingly difficult. The difficulty should not come from coincidence. It should come from the character's own decisions, from the lie that distorts those decisions, and from the people who see through it.
A useful prompt asks the model to escalate the pressure in three specific ways:
> The inciting incident is that Clara's daughter arrives with a clock that belonged to Clara's mother. Give me three escalating complications. One should come from Clara's professional standards, one from a person who notices her avoidance, and one from the clock itself revealing something Clara did not expect.
The model might suggest:
- •Clara insists on restoring the clock to museum standards, but the mechanism is more damaged than it appears, requiring a part she has avoided sourcing since her mother died.
- •The local librarian, who knew Clara's mother, keeps asking questions that imply the clock's history is not what Clara has told herself.
- •While disassembling the clock, Clara finds a hidden compartment containing a letter addressed to her mother from someone she does not recognize.
These complications are good because they all push Clara toward the same confrontation: the belief that she can control loss by controlling objects. The professional complication forces her to handle the very mechanism that connects to her grief. The social complication introduces a witness. The discovery complication brings new information that breaks her private version of the past.
You will almost always have to rewrite the model's suggestions. The mechanism it proposes may be wrong, the librarian may sound too convenient, the letter may feel melodramatic. Treat the suggestions as structural options, not finished scenes. Pick the one that tightens the coil around the character's want.
Designing the Reversal
The reversal is the hardest bone to get right. It must feel both surprising and inevitable. Surprising, because the reader needs a jolt. Inevitable, because the jolt must make sense in retrospect, given everything we know about the character.
The best reversals expose the lie. Clara has told herself that people leave because places like this town fail them. The reversal might be the discovery that her mother chose to stay, that the clock was a gift from someone who loved her, that the daughter did not leave because of the town at all. The reversal is not necessarily a plot twist in the *Twilight Zone* sense. It is a recontextualization of what the character thought they knew.
I prompt the model toward reversals that reframe the want rather than solve it externally:
> Give me three possible reversals for Clara's story. None should involve violence, crime, or a sudden confession. Each should force Clara to realize that her original want was incomplete or mistaken, and each should require her to make a new choice she has been avoiding.
One reversal might be that the clock cannot be repaired at all, and Clara must decide whether to tell her daughter the truth or fabricate a working mechanism the way she has fabricated emotional stability. Another might be that the daughter already knows the truth about the letter and has been waiting for Clara to bring it up. A third might be that the town council's demolition decision arrives during the restoration, forcing Clara to choose between finishing the clock in private and standing with her daughter in public.
The reversal you choose determines the meaning of the story. It is worth spending extra time here, trying several options, and selecting the one that hurts the character in the most productive way.
The Resolution Must Be Decisive
A short story does not have room for a drifting ending. The resolution should answer, in action or in image, the question the inciting incident raised. The answer can be ambiguous, but the ambiguity must be intentional, not evasive.
If the inciting incident was the daughter's arrival with the clock, the resolution should tell us whether Clara can treat her daughter as something more valuable than an object. If the inciting incident was the threatened opera house, the resolution should tell us whether Clara's belief in beauty is strong enough to survive failure. If the inciting incident was the flooded workbench, the resolution should tell us whether grief can be repaired at all.
I often ask the model for a resolution in the form of a final image rather than a summary:
> Give me three possible final images for the story, each under fifty words. Each image should show Clara's hands doing something concrete with a clock, and each should imply whether she has changed or stayed the same.
An image like this is more useful than a paragraph of explanation:
> Clara sets the restored clock on the shop counter beside the untouched suitcase. She winds it. She does not look at the door. The ticking is the only thing she answers.
That image tells us she has retreated again. It is a resolution because it is decisive: she has chosen the clock over the daughter, and the choice has cost her something visible. Whether that ending is tragic or simply honest depends on the story leading up to it.
How AI Helps and Where It Fails
AI is strongest at the beginning of skeleton construction. It can generate a dozen inciting incidents in minutes, suggest complications you might not have considered, and propose reversals that reframe the want. It is weakest at knowing which options belong to your character. The model does not love Clara Wren. It has no investment in her lie. It will produce plausible shapes that have nothing to do with the person you have spent three articles building.
Your job is to reject the options that do not fit and redirect the model. When a suggested reversal feels generic, ask for the fourth or fifth option. When a complication feels externally imposed, ask for one that grows out of Clara's own decision. When an ending sounds like a moral lesson, ask for an image instead.
Think of the model as a brainstorming partner who has read widely but has no taste. Taste is your contribution.
The Skeleton Test
Before you move from skeleton to scene, test the skeleton against five questions.
Does the inciting incident happen to this character and no one else? If anyone could receive the same disruption, the story is not yet specific enough.
Does each complication make the want harder in a different way? If the complications are repetitive, the middle will sag.
Is the reversal both surprising and inevitable? Surprising without inevitability is a trick. Inevitable without surprise is a lecture.
Does the resolution answer the story's real question? The real question is not whether the clock is repaired. It is whether Clara can change.
Can you explain the skeleton in three sentences? If not, it is too complicated for a short story.
If the skeleton passes all five tests, you are ready to write the first scene. If it fails any test, return to the model and ask for alternatives until it passes.
From Skeleton to Scene
The skeleton tells you what must happen. The scene tells you how it happens. The transition from one to the other is where most AI-assisted drafts go wrong, because the model will happily write a competent scene from a weak skeleton, and the competence will hide the weakness.
Do not let the model write the scene until the skeleton is sound. Once it is, you can prompt the model for a first draft of the opening scene with the skeleton embedded in the prompt:
> Write a 600-word opening scene for a short story. The scene begins the moment Clara Wren's estranged daughter enters the clock shop with a damaged clock. Clara wants to treat the visit as a repair transaction. Her daughter wants something Clara is not yet willing to give. Use Clara's verbal tic of speaking in mechanical metaphors and her habit of dropping the subject in emotional sentences. Do not explain the backstory. Let it leak through behavior and object.
The scene the model produces will be a starting point, not a final draft. You will rewrite every line that sounds like the model and keep every line that sounds like Clara. The skeleton ensures that even a mechanically competent draft has a spine.
The Temptation to Fill the Skeleton Too Early
A common mistake is to start writing scenes before the skeleton is complete. You have a good inciting incident, so you begin there. You write two pages, then realize you do not know what the reversal is. The draft stalls. You ask the model to continue, and it invents a plot twist that does not belong to the character.
Resist this. The skeleton is not bureaucracy. It is the cheapest place to make mistakes. A bad reversal on the page costs you a draft. A bad reversal in a bullet point costs you thirty seconds.
I usually spend at least as long on the skeleton as I do on the first draft. If the skeleton is right, the draft writes itself in surprising ways. If the skeleton is wrong, no amount of good prose will save the story.
Plot Skeletons for Non-Fiction Writers
Even if you do not write fiction, the skeleton exercise is valuable. An essay has an inciting incident: the question or provocation that opens it. It has rising action: the argument that builds pressure on the initial claim. It has a reversal: the moment when a counterexample or deeper framework reframes the argument. It has a resolution: the final claim or image that answers the opening question.
The same structural discipline applies. The difference is that the character whose want drives the plot is often the reader, or the writer, or the subject under examination. Learning to build a skeleton for a short story makes you better at building one for an essay, a profile, a longform argument, or a reported feature.
For Next Time
On Wednesday we will look at "Short Story — Plot: The Three-Act Machine," where we take the skeleton and fit it into the classic three-act structure. We will explore how the same bones can be arranged to produce different emotional rhythms, and how AI can help you test those arrangements before you commit to a full draft. The character you built last week is finally about to move.
Your homework until then: take the character you have developed and build a four-bone skeleton for a short story. Write each bone in a single sentence. Then ask the model to suggest three alternative reversals. Choose the reversal that most painfully recontextualizes your character's lie. Do not write the scene yet. If the skeleton is sound, the scene will demand to be written.
A plot skeleton is not a formula. It is a promise. It promises the reader that the person they have agreed to follow will be tested, that the test will matter, and that by the end something will have changed. Build the bones well. The rest is muscle and nerve.
---
*Harry Mercury, Editor in Chief* *The SMF Works Project* *Week 9, Article 1*