| name | crafting-stories |
| description | Craft, develop, diagnose, or revise narrative stories using the principles of dramatic structure and character. Use this skill whenever the user wants help writing a story, novel, screenplay, short story, or narrative scene; developing a plot, premise, or logline; building characters or character arcs; fixing a draft that "feels flat," "drags," or "isn't working"; outlining or structuring a narrative; writing a scene that turns; or critiquing/editing fiction. Trigger this even when the user just says "help me write a story about X," "my second act sags," "this character is boring," "how do I end this," or pastes a draft and asks for feedback — narrative craft problems are rarely solved by raw prose generation alone, and this skill supplies the diagnostic framework. |
Crafting Stories
A working craft system for inventing, developing, diagnosing, and revising stories. It is built on the dramatic tradition that runs from Aristotle's Poetics through modern story theory: a story is not "stuff that happens" — it is a designed sequence of meaningful change, driven by conflict, revealed through choice under pressure.
Use this skill to help a writer at any stage: a blank page, a vague premise, a sagging middle, a flat character, or a finished draft that "doesn't work" for reasons they can't name.
The one idea everything hangs on
A story is the honest account of how and why a human being's life changes, told through the gap between expectation and result.
A character wants something. They take an action expecting the world to cooperate. Instead the world responds with a reaction that is more powerful, surprising, or contradictory than they expected — opening a gap between what they thought would happen and what actually did. To close that gap they must take a riskier action against stronger resistance. This gap, repeated and escalated, is the engine of every compelling story. When you don't know why a scene is dead, the gap is almost always missing: the character got exactly what they expected, so nothing turned.
Hold this next to a second idea:
Meaning comes from change in a value. A story event takes a charged human value — life/death, love/hate, freedom/captivity, truth/lie, hope/despair, justice/injustice, maturity/naivety, connection/isolation — and shifts it from one pole to the other. If nothing of value changes, no story event has occurred, no matter how much motion is on the page.
Everything else in this skill is machinery for producing and sustaining those two things: the gap and the value change.
How to use this skill
First, locate where the writer is. Don't run the whole framework at them; meet the actual need.
| The writer says... | Go to |
|---|
| "Help me come up with / develop a story" | Workflow A: Generate (below) |
| "Here's my draft / scene / outline — what's wrong?" | Workflow B: Diagnose (below) + references/diagnosis.md |
| "How do I structure this / my middle sags / how do I end it?" | references/structure.md |
| "This scene is flat / how do I write a scene that works?" | references/scene-craft.md |
| "My character is boring / how do I build an arc?" | references/character.md |
| "What's this story actually about / what's the theme?" | references/meaning.md |
Read a reference file before leaning on it — each one carries detail this page only summarizes. Then return here for the working principles below, which apply across all of them.
Workflow A: Generate a story from scratch
Don't start with plot. Start with the engine, then grow structure outward from it.
- Find the value at stake. What dimension of human life will this story move, and in which direction? "A man moves from isolation to connection." "A woman moves from comfortable lie to costly truth." This is the spine before there is any plot.
- Find the desire and the resistance. What does the protagonist consciously want (the object of desire)? What stands against it — and crucially, is the strongest resistance inside them, in their relationships, or in the wider world? The best stories braid all three (see
references/character.md on the three levels of conflict). A protagonist is only as deep as the forces opposing them are strong.
- Find the inciting incident. What single event knocks the protagonist's life out of balance and makes the desire urgent and unavoidable? It should feel both surprising and, in retrospect, inevitable. Everything before it is setup; the story proper begins here.
- Find the worst possible turn (the crisis). Before you outline the middle, imagine the moment late in the story where the protagonist must make the hardest choice — between two things they want but can't both have, or between two bad options. This crisis decision is where character is most deeply revealed. Knowing it tells you what the whole middle must build toward.
- Now grow the middle backward and forward as a chain of escalating gaps (see
references/structure.md). Each major turn should raise the stakes and narrow the protagonist's options until only the crisis is left.
Offer the writer choices at each step rather than dictating; storytelling is theirs to own. But push for specificity — vague desire and vague resistance produce vague stories.
Workflow B: Diagnose a draft
When given a draft, don't line-edit prose first. Find the structural failure, because prose polish on a broken structure is wasted. Run the diagnostic questions in references/diagnosis.md. The most common root causes, in rough order of frequency:
- No gap. Characters get what they expect; scenes resolve without surprise. → The world must push back harder than anticipated.
- No value change. Scenes convey information or "develop the relationship" but nothing flips from positive to negative or back. → Make each scene turn.
- Weak antagonism. The opposing forces are too easily overcome, so nothing is proven. → Strengthen what stands in the way until the desired thing is genuinely hard-won.
- A passive protagonist. Things happen to them; they don't drive. → Give them desire and force them to choose and act.
- On-the-nose dialogue. Characters say exactly what they feel and mean. → Push the real meaning into subtext (see
references/scene-craft.md).
- Exposition dumped, not dramatized. Backstory delivered in lumps. → Convert exposition into ammunition characters use against each other inside conflict.
- A crisis with no real dilemma. The "hard choice" was actually easy. → Make both horns of the choice cost something real.
Working principles (apply everywhere)
Show choice under pressure. Character is not what a person says about themselves, nor their quirks and surface traits (that's characterization) — it's revealed by the choices they make when the stakes are highest. The greater the pressure, the deeper and truer the revelation. A choice between right and wrong is trivial; a choice between two rights, or two wrongs, exposes who someone really is.
Conflict is non-negotiable. Nothing in a story moves forward except through conflict. "Conflict" doesn't mean shouting or violence — it means the friction between desire and resistance. A quiet scene of two people not quite saying what they mean is full of conflict. A loud car chase with no value at stake is empty.
Every scene must turn. A scene that ends on the same value charge it began on is a non-event, however well written. Open on one charge (the character is, say, hopeful/positive about getting the job) and end on the opposite (humiliated/negative) — or the reverse. If it doesn't turn, cut it or fold its essential content into a scene that does.
Earn the ending; don't engineer it. The climax must be the inevitable-feeling result of everything the protagonist chose, not a coincidence or a rescue from outside. Coincidence can launch a story but must not resolve it. The audience will forgive almost anything except a cheated ending.
Surprise, then satisfy. The best turns are surprising in the moment and inevitable in hindsight. If a turn is predictable, it's flat; if it's surprising but unprepared, it feels like a cheat. Plant the cause early and quietly so the effect lands as both shock and recognition.
Less is more, and late is better than early. Enter every scene as late as possible and leave as early as possible. Withhold information to create curiosity; reveal it through conflict, not announcement. Trust the reader to do work.
Write the truth of human behavior. Genre conventions and structure are scaffolding, not the building. The thing readers actually come for is insight into how people really are under pressure. Honesty about behavior is what separates a competent story from a moving one.
A note on register and genre
These principles are load-bearing across forms — literary fiction, genre fiction, screenplay, short story, even narrative nonfiction — but their application varies. A minimalist literary story may soften the inciting incident and leave the ending open; a thriller leans hard on escalating external antagonism; a character study lives in inner conflict. Calibrate to the writer's intended form and ambition rather than forcing every story into one mold. references/structure.md covers the spectrum from tightly-plotted classical design to open, anti-structural forms, and when each serves.
What good help looks like
Be a perceptive development editor, not a rule enforcer. Name what's working before what isn't. Diagnose with the craft vocabulary above so the writer learns to see their own work more clearly. Offer concrete options and rewrites, not just notes. And remember the goal is never "follows the structure" — it's a story that makes a reader feel something true.
