Narrative Review

Assess how well a story delivers the experience it promises and identify writing choices that weaken it. Base the judgment on the text, the author's purpose, and relevant conventions. A trope or a departure from a template is not inherently a flaw. These criteria guide editorial judgment; they cannot guarantee every reader's response.

Scope and Expectations

Identify the available material: a complete work, unfinished draft, synopsis, chapter, scene, or interactive route. Record the medium, genre, audience, intent, chosen structure, and constraints when provided. If they are missing, make a brief hypothesis from the text and continue. When the premise is unclear, the test "when X happens, Y must do Z before W, but faces V" exposes the gap without requiring that wording. Ask for the text when there is no material to review.

Read the entire submitted unit before giving a diagnosis. For long works, proceed in sections and maintain a compact record of scenes, chronology, rules, character knowledge, promises, and payoffs. Verify final findings against the original passages. If only a sample was read, state its scope and present the diagnosis as partial.

Distinguish three sources of expectations:

• The author's stated intent and conditions.
• Promises established by the work through its opening, tone, questions, and emphasis.
• Relevant conventions of its genre or medium, noting when they are inferred.

A fragment does not establish that a resolution is missing or that a clue is never paid off. A synopsis supports architectural analysis, but rarely proves problems with dialogue, style, or reading pace. Mark those areas as not assessable. Respect spoiler limits. If they prevent an explanation, identify the limitation without revealing restricted content.

Structural Reading

Reconstruct the causal sequence and compare it with the order in which events are presented. Locate the story engine, decisive changes, open questions, and the effect of the ending when one is available.

When the story resolves a conflict, audit the ending against five conditions: the solution uses abilities, relationships, objects, or knowledge that were prepared; it arises from a decision; it respects the established rules; it exacts a price; and it answers the central question. A failed condition is a lead to verify against the setup, not a verdict by itself.

When the author requests a particular model, assess how the story performs the model's functions. Otherwise, select only the lenses that clarify the work, using the table below. Do not impose fixed percentages, a mentor, a personal antagonist, positive transformation, or an action climax on every story. Structure regulates changes, not exact percentages.

An absence of explicit conflict can suit a contemplative story. In that case, assess development, contrast, discovery, relationships among images, and changes in perception. An episodic narrative can achieve unity through voice, setting, or theme. Strict causality carries different weight in different kinds of work.

Planning methods such as the Snowflake Method, the Story Spine, or the Lester Dent formula are not requirements a manuscript must satisfy. Propp, Polti, Booker's seven plots, and computational emotional arcs are descriptive lenses. They do not establish quality or cultural completeness.

Selecting Analytical Lenses

Evaluation Areas

Review the relevant areas, even when the report develops only the significant ones:

• Promise and closure: central question, genre experience, expectations created, payoffs, and consequences. Distinguish an episode's closure from a series' closure.
• Structure and causality: inciting event, shifts in strategy or understanding, progression, crisis, climax, and denouement where applicable; continuity of time, space, and rules. A setting is active when it limits options, creates opportunities, or changes the cost of acting. A subplot must touch the main plot through causality, decision, or meaning; one that disappears without altering anything is probably another story.
• Characters and relationships: desire, motivation, limits, individual or collective agency, opposition, decisions, cost, reactions, and development. The opposition may be a person, institution, community, environment, clock, secret, illness, or inner contradiction; it works best when it pursues a goal incompatible with the protagonist's and has the means to reach it, and the deepest conflict sets two legitimate values against each other. Consider flat, negative, disillusionment, and ambiguous arcs.
• Information and twists: setup, clues, false interpretations, point of view, available knowledge, and the consequences of revelations.
• Scenes and pacing: function, change or deepening, exposition, repetition, ellipsis, pauses, and contrast. Do not require a complete conflict in every scene, but even a quiet exchange has someone who wants something.
• Voice, dialogue, and theme: focalization, distinct voices, intention, subtext, tone, images, motifs, and the relationship between what the work states and what it shows. Not every line needs subtext; a direct confession can be the hardest and most dramatic act in a scene. In dialogue, check that each line pursues something, that physical action anchors the exchange, that tags stay plain unless a variation adds meaning, that emotion shows in behavior before the narrator names it, and that quick exchanges alternate with pauses. Motifs and symbols work best when they change with context and stay unexplained.
• Problematic devices: clichés, arbitrary solutions or misfortunes, plot armor, inconsistencies, information manipulation, and emotional shortcuts. Apply the distinctions below before assigning a label.

Promises and Twists

Record each major promise through these states: introduction, expectation created, later setup, payoff, or pending. Strong emphasis creates a greater expectation than an incidental mention. Chekhov's gun allows atmospheric, character, and thematic functions; it does not require every object to reappear in the action.

A twist works when it reinterprets rather than merely adds information, was set up with visible clues or rules, changes decisions and consequences, and respects causality and point of view. The test is double: the first reading should surprise; the second should show that the story never cheated. Compare the initial interpretation with the revealed truth. Look for facts compatible with both, a plausible basis for the mistaken interpretation, at least one verifiable clue, and subsequent consequences. A red herring can mislead through interpretation without falsifying facts. Check what each character knew and what the narrative voice could disclose.

Treat these forms as weak unless the text supports them: "it was all a dream" without consequences or setup; a twin, clone, or secret identity introduced in the resolution; a new antagonist replacing the developed conflict late; information the audience could not know in a mystery presented as solvable; a resurrection that invalidates earlier grief and cost; a betrayal motivated only by surprise; a twist that contradicts established rules of the world.

An unreliable narrator may lie. Assess the consistency of the voice, signs of its limitations, and the expectations established by the story. Do not declare a cheat merely because the narrator's account is false. In a mystery that invites the audience to solve the puzzle, check whether the necessary information was available. Other forms of mystery may promise uncertainty or discovery without offering a deduction game.

Information and Tension

Identify which mode a passage relies on: suspense, where the audience knows enough about the danger and waits for the outcome; surprise, where an event breaks the expectation at once; mystery, where a cause or identity is unknown and the audience reconstructs it; or curiosity, where an anomaly opens a less urgent but persistent question. A story may alternate them. Hiding everything does not create suspense; it often removes it. Dramatic irony works when the audience knows something a character does not, and the tension comes from when, how, and at what cost they will find out. A MacGuffin moves the action; it does not replace the emotional conflict. A cliffhanger earns its interruption when the next unit soon satisfies or transforms the question; repeated false dangers erode trust.

Where the work seeks tension, it requires an uncertain outcome, something valuable at stake, forces pushing in incompatible directions, enough information to anticipate consequences, and delay with progression rather than mere postponement. When a passage feels flat, find which of these is missing before prescribing more incidents. Accelerating means compressing what matters less and bringing decisions closer, not only shortening sentences.

Coincidences and External Interventions

Deus ex machina: a decisive solution depends on a force, rule, or ability that lacks adequate setup. Check prior events and the protagonist's participation before applying the label. Outside help may result from an earned alliance, an earlier request, or collective action.

Diabolus ex machina: an informal term for an intervention that artificially worsens the situation or imposes defeat. Distinguish unexpected harm from arbitrariness. An early disaster can establish the premise without prior foreshadowing. A later complication can arise from the established world or from earlier choices. Such cases may work as a catalyst or peripeteia without constituting a flaw.

The risk increases when a late intervention invalidates previous efforts, gives the antagonist a new power, or prolongs the conflict without setup. The misfortune may be unforeseeable for the characters, but it should be coherent for the audience on rereading. Consider whether absurdity, chance, or powerlessness belongs to the story's design, as in cosmic horror, tragedy, satire, or dark comedy, and whether its execution supports that design. A tragedy is not flawed merely because it ends badly.

Invented examples for calibrating the criterion:

• A storm sinks the ship in the opening and begins the survival story: a plausible catalyst with no automatic need for foreshadowing.
• During the escape, the hero discovers that the enemy can stop time; nothing earlier allowed for this ability: a potentially arbitrary misfortune. Verify the rules before reaching a conclusion.
• The creature that prevents the escape left tracks earlier, and the chosen route crosses its territory: a prepared complication. Assess the decisions and consequences.

Devices That Require Contextual Diagnosis

Verifying Findings

For each potential problem:

1. Locate the passage that produces it and any relevant setup. Use chapter and scene, verified line numbers, or a short identifiable quotation. Do not invent page numbers or quotations.
2. Explain which expectation or rule is affected and how that may shape the reading experience.
3. Look for a supported alternative explanation: earlier setup, an unreliable narrator, irony, ellipsis, genre intent, or information outside the submitted fragment.
4. Decide whether it is a supported flaw, an interpretive risk, or an editorial preference. If the evidence is insufficient, discard it as a flaw or leave one concrete question.
5. Propose the smallest intervention that addresses the cause. Explain its cost or side effect when it changes the ending, theme, voice, or relationships.

Do not present a hypothesis about the author or audience as fact. Describe reading effects as reasoned interpretations. Group symptoms that share a cause while preserving their separate locations; do not count the same problem more than once.

Prioritize by impact:

• High: compromises the central conflict, a decisive rule, comprehension, or the main payoff.
• Medium: appreciably weakens a sequence, relationship, revelation, or progression.
• Low: a localized problem whose correction improves clarity, economy, or nuance.

Keep priority separate from confidence: a serious suspicion with little evidence remains a suspicion. Do not invent findings to fill a quota. A story may work without requiring corrections.

Diagnostic Checklist

Run through these questions before closing the diagnosis. A "no" is a lead to verify, not a finding by itself.

• Global structure: Does the story promise a recognizable experience? Does the inciting event genuinely alter the protagonist's life? Do the scenes form a causal chain? Does the midpoint change knowledge, strategy, or power? Does the crisis offer options with a cost? Does the climax answer the central question through action? Does the ending show sufficient consequences?
• Characters: Does the protagonist have desire, motivation, and the capacity to decide? Does the opposition pursue its own goal? Do secondary characters want something beyond helping or hindering? Is the arc demonstrated under pressure? Do relationships change because of what happens?
• Information: Do the necessary rules appear before the conflict is resolved? Do the clues admit both the truth and the mistaken interpretation? Does the point of view avoid hiding thoughts artificially? Does exposition arrive when it changes a decision or a reading? Are deliberate mysteries distinguishable from logical gaps?
• Scenes and pacing: Does each scene have a clear objective or question? Does its outcome change the plan, the stakes, or a relationship? Is there contrast between intensity and calm? Do characters react to losses and revelations? Do repetitions raise pressure or meaning?
• Originality and coherence: Do tropes carry their own details and consequences? Is the main conflict resolved by something other than coincidence? Do the rules also apply to the protagonist? Does the opposing force keep its strength and intelligence through the ending? Does the theme emerge from actions rather than speeches alone?

Report

Write the report in the user's language unless requested otherwise. Match its length to the material and assignment. For a general review, include:

1. Verdict and scope. State whether the work fulfills, partly fulfills, or compromises its promises. Use "not assessable from the available material" when appropriate. Identify inferred expectations and the material reviewed.
2. Strengths to preserve. Identify effective mechanisms with brief evidence, especially those that a correction might damage.
3. Status by area. A short table with the area, status, and basis can distinguish "meets expectations," "partial," "supported problem," "not applicable," and "not assessable." Do not use numerical averages unless the user requests them.
4. Prioritized findings. For each high or medium finding, provide an identifier, priority, confidence, location, evidence, mechanism, impact, and concrete proposal. A low finding may be compact: identifier, location, evidence, and proposal. Distinguish optional recommendations from necessary corrections.
5. Revision plan. Order interventions by dependency: architecture and causality, characters and information, then scenes and language. For each major change, state how to verify that it works and what should be preserved.

Spelling, grammar, punctuation, and typography are not narrative findings. When the text has such errors, say so in one line at the end of the report and point to the proofread skill, which runs once the text is final. Do not provide a lengthy plot summary unless it is needed to explain a finding. For focused assignments, develop only the requested area and the dependencies that affect it.

A review authorizes diagnosis and recommendations. Modify the manuscript only when the user requests corrections or rewriting, following the rules under Applying Corrections. You may include a short example labeled as a proposal when it clarifies an intervention; never present it as a quotation from the original. For a second review, verify the findings against the new version and identify which are resolved, remain, or have changed.

Verifiable Corrections

Prefer a concrete intervention to "add depth" or "improve the pacing": establish a limitation before the climax, show the choice that leads to a loss, move an explanation to the point where it becomes necessary, or preserve a reaction after a revelation. To renew a trope rather than remove it, name the satisfaction it promises, keep that satisfaction, and change the causes, setting, relative power, or cost; give each character an agenda independent of the trope and follow the consequences past the expected moment. An inversion made only for surprise is as predictable as the original.

Explain which passage should change, what effect the change should produce, and how to verify it on rereading. If a proposal requires new facts, identify them as options. Check that solving one problem does not remove the surprise, alter the theme, or make an earlier or later scene inconsistent.

Applying Corrections

Rewrite only when the user asks for it, and only the findings the user selected. Where a finding offered options, have the user choose before touching the text.

Edit only the passages cited in those findings, including the place where a proposal says the fix goes. In this first stage everything else stays byte for byte, line endings included. Do not add scenes unless the chosen option requires one. An inserted line imitates its neighbors: sentence length, vocabulary, tense, focalization, and register. It shows what the finding needs and does not explain it, and it carries no filler connectives or "not X but Y" frames. When it plants a rule or a clue, its factual content is fixed: the wording may change, the meaning may not.

If the humanize skill is available, run it in two places. First on the inserted lines, supplying the rest of the text as protected material so it can read the author's voice and imitate it. Then, once the findings are applied, run it over the whole text as a second stage of the correction: prose written by a person can still be stiff, redundant, or unclear. Give humanize a style instruction built from the review: the genre and register, the strengths to preserve the report named, and the deliberate devices that are style rather than defects, such as anaphora and parallel negations, contrasts of the form "it was not X; it was Y", deliberate fragments and one-word paragraphs, escalation by repetition of a word, and, in Spanish, the raya that opens dialogue or sets off a narrator's inciso and the ellipsis character …. Tell humanize that the style instruction overrides its own defaults on these points. Supply the spoken words of dialogue, including dialogue the correction inserted, and any quoted material as protected text. Leave headings, epigraphs, the author's notes, and documents quoted inside the story, such as a letter or a diary entry, outside the pass. The pass may change rhythm, word choice, and clarity; it never changes facts, names, events, their order, or what a character says, and it keeps the file's line endings. After it, check that every device named in the style instruction survived, and revert any edit that removed one. When humanize is not available, say so in the change log. Both passes run by default. Skip them only when the user explicitly asks, for instance by saying not to humanize or that humanize will run later in their own workflow; never infer it from the size of the correction or from context. An explicit request skips both passes, the inserted lines included: the inserted lines then rely on imitating their neighbors, as the drafting rule above requires. When humanize is skipped, say in the change log that it was skipped at the user's request, and write "none" with that reason in its humanize part. When the user asks to change little, for instance by saying not to change much, the passes still run but are limited to errors and clarity problems, such as contradictions, broken constructions, and obvious redundancies; they leave rhythm and style alone, and the change log says the limit was applied.

Deliver the complete corrected text and a change log in two parts: one entry per finding with identifier, location, text before, text after, and what to verify on rereading; then one entry per edit the humanize pass made: location, text before, and text after, so each can be reverted on its own. Give every location in the original's numbering, and locate an insertion as "after Lnn". Then review the new version as a second review: state which of the applied findings are resolved, remain, or changed, mark the findings the user did not select as not selected, and confirm that everything listed under strengths to preserve is intact.

Example Use

"Use /narrative-review to review this complete short story. Assess whether it fulfills its mystery promise and whether the final twist is properly set up. Prioritize the problems, cite the scenes, and propose corrections that preserve the ending."