The direct answer, stated without decoration
A synopsis is a one-page summary of your entire plot, written in present tense, with the ending included. It exists for agents and editors, not for readers. You write it by stating your protagonist and their goal, the obstacle blocking them, what failure costs them, and how the story actually resolves — then compressing those five facts into three paragraphs called setup, complication, and climax. Most agents ask for about 500 words, which means roughly one page. Enter those five beats into the tool above and it arranges them into that three-paragraph shape, counts the words against your target, confirms the ending is present, and flags past-tense verbs that break submission format.
That paragraph is the whole method. Everything below explains each decision inside it, shows the arithmetic the tool performs on your words, walks through hypothetical drafts that succeed and fail, and covers how the rules bend by genre.
Why agents ask for a synopsis at all
An agent reads a query letter to judge the voice, sample pages to judge whether the prose sustains it, and the synopsis to judge whether the plot holds together across a full book. Many writers assume the synopsis repeats the query at greater length, so they submit suspenseful summaries hiding the resolution. That wastes the document: the agent already assesses suspense from the pages, while only the synopsis can prove architecture — an escalating middle, personal stakes, an ending that pays for the beginning.
Read the synopsis as a structural X-ray. When an agency requests pages plus a synopsis, the second document means: show me the experience, then prove the engineering. Writers who absorb this stop trying to make the synopsis thrilling and start making it legible.
A synopsis is not a blurb, and confusing them sinks both
The back-cover blurb sells a book to a reader. It introduces the hero, sketches the hook, and stops before the resolution so the reader must buy the book to learn the outcome. The submission synopsis briefs an agent on the complete plot. It introduces the hero, tracks every major turn, and states the ending outright, because the agent must evaluate the story before offering representation. One document withholds the ending deliberately. The other is worthless without it. They share subject matter and nothing else. Blurb copy uses fragments and teasing questions because its task is desire, while synopsis prose uses plain declarative sentences because its task is clarity. Writers who import blurb habits into a synopsis force the agent to guess at the plot instead of judging it.
The tool on this page builds the second document, never the first. It takes your five beats and assembles a full-plot submission synopsis with the ending stated, then checks length and tense against agent expectations. It does not draft teaser copy, invent taglines, or hide your resolution to manufacture suspense. If you need back-cover text for readers, treat that as a separate assignment with opposite rules. Keep the two documents in separate files, compose them in separate sittings, and never submit one where the other was requested.
What this tool actually does with your words
First, the honest boundary: this tool does not write your synopsis, and it does not improve your sentences. It performs deterministic assembly and measurement on text you supply. You type five short inputs — protagonist, want, obstacle, stakes, ending — plus a genre and a word target. The tool slots those inputs into three fixed paragraph templates, counts the words in each paragraph and overall, expresses the total as a percentage of your target, scans every word against two fixed verb lists to judge tense, and checks whether the ending field holds anything at all. Nothing leaves your browser, no outside service is consulted, and no new prose is generated beyond joining your phrases with a small amount of fixed scaffolding.
That limit is the point. A synopsis stands or falls on plot decisions only you can make: whose story this is, what they want, what blocks them at rising cost, and what the resolution demands. No software can derive those answers from a premise. What software can do reliably is enforce format: verify the ending exists, measure length against the agent's number, and catch past-tense drift. Those three checks are exactly what this page automates.
Treat the tool as a format gate, not an author. Draft your five beats in your own language first, run them through the assembly, and revise the beats themselves — not the tool settings — until every flag clears.
How the assembly math works, step by step
The tool builds three paragraphs from your five inputs with templates that never vary. Paragraph one, the setup, joins your protagonist phrase to your want phrase with the single added verb wants. Enter Mara Quinn, a disgraced cartographer as protagonist and to earn a pardon by mapping the Closed Coast as the want, and the setup becomes Mara Quinn, a disgraced cartographer wants to earn a pardon by mapping the Closed Coast. Fifteen words, one sentence, hero plus goal, no ornament. The template capitalizes the opening letter and closes with a period; everything else is your wording.
Paragraph two, the complication, turns your obstacle and stakes inputs into two sentences by capitalizing each and adding terminal periods. A rival surveyor burns every map she finishes and Her brother faces exile if she fails again pass through otherwise untouched. Sixteen words across both sentences. Nothing is reordered, nothing is rephrased, and no connective tissue is invented. The paragraph functions only if your two inputs genuinely escalate — an external block plus a personal cost — because the template contributes no escalation of its own.
Paragraph three, the climax, converts your ending input into one sentence the same way. She burns her pardon to save the hidden villages becomes the entire third paragraph. Nine words. Leave the ending field empty and the third paragraph stays empty, the spoiler check fails, and the verdict reports the failure instead of masking it.
Word counting uses the simplest defensible rule: split the assembled text on whitespace and count the pieces. The tool counts each paragraph separately for the table, then totals them for the metrics. The percentage beside your target is the total divided by the target, multiplied by one hundred and rounded. A 40-word draft against a 50-word target reads as 80 percent; the same 40 words against 500 read as 8 percent.
Tense scoring tokenizes the assembled draft into lowercase words and tests each against two fixed lists: past forms such as was, had, burned, traded, failed, and decided, and present forms such as wants, burns, faces, fails, saves, and will. Past strictly greater than present means past-leaning; present above zero with past at or below present means present-leaning.
Worked numbers: one practice draft measured twice
Take the default beats: a fifteen-word setup, a sixteen-word complication, and a nine-word climax, forty words in total. Run that draft against a 50-word practice target. Forty divided by fifty is 0.8, times one hundred is 80 percent, which lands inside the tool's on-target band — 60 percent at the floor, 200 percent at the ceiling. The ending field carries a complete resolution, so the spoiler check passes. The verbs wants, burns, finishes, faces, fails, and saves all match the present-tense list, with zero past-tense matches, so the tense signal reads present-leaning. Every gate clears and the verdict passes.
Run the same forty words against the standard 500-word submission target and the math flips: 8 percent, far beneath the 60 percent floor. The ending remains present and the verbs remain present-tense, yet length alone triggers a warning prescribing the repair: add the midpoint reversal and the dark moment as one sentence each. The tool never judges plot in isolation; it judges the fit between a summary and a stated requirement, which mirrors how an agent reads.
Note the band edges. Sixty percent of 500 is 300 words and 200 percent is 1000, so the band tolerates real variation between agents. Set the target to the precise number your agent names.
What each paragraph must carry, and why three suffice
Three paragraphs sound impossibly tight for a whole novel until every paragraph receives a distinct assignment. The setup answers who wants what: a named protagonist with one identifying trait and a concrete goal with a visible finish line. Misty goals such as finding herself collapse here, because a synopsis must state plot, and plot demands an outcome the ending can confirm or deny.
The complication answers what blocks them and what failure costs: an obstacle that acts — a rival who burns maps, a detective who narrows suspicion — plus stakes fastened to a specific person. The brother facing exile is why Mara cannot quit when the maps burn. Without that sentence, the setup describes employment, not story.
The climax answers what they finally do and what it costs: the decisive action stated as fact — she burns her pardon — plus the party the action serves — to save the hidden villages. An ending reporting sacrifice for a named cause lets the agent feel the book cohere. All three paragraphs together contain no theme statements, no comparisons to published authors, and no rhetorical questions. Those belong in the query letter.
Illustrative hypothetical example one: the cartographer passes
Picture a hypothetical writer preparing a fantasy submission — an illustrative scenario, not a real manuscript. She enters Mara Quinn, a disgraced cartographer as protagonist, to earn a pardon by mapping the Closed Coast as the want, A rival surveyor burns every map she finishes as the obstacle, Her brother faces exile if she fails again as the stakes, and She burns her pardon to save the hidden villages as the ending. She selects fantasy and sets a 50-word practice target while learning the form.
The assembly totals forty words, which is 80 percent of the practice target. The ending sentence names the final action and its price, so the spoiler check passes, and every verb lands on the present-tense list. The verdict passes, and the table displays the healthy shape: a lean setup, a heavier complication, a brief climax that lands like a closing door.
Her next move is proportional expansion. Each paragraph needs roughly ten times its length to reach a 500-word submission version, and the growth must add plot rather than adjectives: the inciting commission and pardon terms in the setup, a midpoint discovery and an early dark moment in the complication, the physical mechanics of the choice in the climax. She composes those sentences herself, returns the expanded beats to the five fields, and reruns the tool against 500 until the verdict passes again.
Illustrative hypothetical example two: the archivist fails, then repairs it
Now picture a second hypothetical writer preparing a mystery — again illustrative, not a real case. He enters Dev Okafor, a night-shift archivist, to clear his name after a reading-room theft, A detective treats him as the only suspect, His sister loses her library apprenticeship if he runs, and leaves the ending field blank to preserve the surprise. He sets a 500-word target.
The verdict fails, naming the precise cause: no ending detected. The writer imagined the agent as a reader to tease. The agent is instead an evaluator who must confirm the mystery resolves fairly — culprit identified, clue chain complete — before requesting pages.
He revises with one ending sentence: He exposes the head librarian with the misfiled accession log and accepts dismissal to protect his sister's placement. Suppose his three paragraphs now total 46 words against the 500 target — 9 percent, still short. The tool directs him to add the midpoint reversal and the dark moment: the detective's star witness recants at the midpoint, and the apprenticeship hearing moves forward. Each rerun returns fresh numbers until the count clears 300 and the tense signal holds.
How the tense check works and why agents insist on present tense
Submission synopses use present tense by convention even when the novel runs in past tense: present tense keeps plot summary distinct from narrative voice, so the agent weighs structure without being swayed by style. A past-tense synopsis reads like an excerpt, inviting the wrong judgment.
The tool encodes the convention as arithmetic, not grammatical analysis. It lowercases the assembled draft, splits it into tokens, and counts matches against its two verb lists. A draft reading was exiled and had decided gathers past hits quickly. A draft reading faces exile and decides gathers present hits. When past strictly exceeds present, the verdict warns and prints both counts so the margin is visible.
The lists cannot capture every verb in English, and this page claims nothing of the kind. Rare verbs and participles may escape both lists, which is why the metric is labeled a signal rather than a grade. Its assignment is catching the dominant failure — a draft composed in the manuscript's narrative past — and that failure invariably brings common verbs such as was, had, and decided along with it. When your draft warns, convert verbs clause by clause beginning with those three. Rerun after each pass and watch the counts cross.
One trap deserves attention. The template contributes wants to every setup paragraph, so each draft opens with one present-tense token regardless of your input. Never treat a narrow escape as approval. Read your own verbs in isolation and convert every narrative verb to present tense by hand before submitting.
The spoiler check: one rule with no exceptions
The spoiler check is the simplest computation here and the strictest rule in the process. If the ending field holds three or more characters, the ending counts as present. Anything shorter counts as missing, the climax paragraph renders empty, and the verdict fails. No partial credit exists for suggestive closings, no allowance for endings implied by the stakes, and no genre exemption. A romance closing on love triumphs and a thriller closing on justice prevails both fail, because neither states the decisive action or its cost.
Writers resist this rule for three reasons, each answerable. They fear the book will seem predictable once the ending is known — yet agents assess predictability from construction, not surprise. They worry the ending sounds flat in one sentence — yet flat summary endings routinely belong to vivid novels. And they were told never to spoil, by advice written about readers and blurbs that does not transfer. The submission synopsis is a professional briefing, and briefings omit nothing.
Compose the ending sentence to a fixed pattern: protagonist takes decisive action at a cost to serve a named beneficiary. She burns her pardon to save the hidden villages fits exactly — action, cost, beneficiary.
Length bands: what short, on-target, and long each signify
A short verdict — under 60 percent of target — almost always signals missing middle rather than economical prose. The structural prescription: add the midpoint reversal, the event flipping the protagonist's fortune halfway through, and the dark moment, the point of maximum cost before the climax. One sentence each. If your draft runs short and you cannot name your midpoint, stop editing the synopsis and outline the manuscript instead.
An on-target verdict certifies proportion against the requested length, nothing beyond that. Never mistake the pass for an agent's endorsement. After the tool approves your draft, verify every named character alters the ending or remove them, and verify the final sentence states the final choice instead of trailing into aftermath. The tool validates the container. You validate the contents.
A long verdict — over 200 percent of target — diagnoses subplot sprawl. Delete in a fixed order. First remove backstory older than the inciting incident. Then remove every subplot whose absence leaves the ending intact. Finally merge minor figures into functions — the rival, the ally, the authority — until at most four names remain.
Genre-specific guidance for the five beats
Fantasy submissions stand or fall on cost. State the price of magic inside the complication paragraph, before the climax spends it, and restrict worldbuilding to what the plot consumes: the coast, the pardon, the rival, the brother. Endings that spend the goal itself — burning the pardon to rescue others — mirror the genre contract of power surrendered for people.
Mystery submissions must do what the genre forbids elsewhere: identify the culprit, state the decisive clue, and explain the false suspicion plainly. Place the false suspicion in the obstacle field, the personal pressure in the stakes field, and the true solution with its mechanism in the ending field. If exposure makes your mystery feel thin, the manuscript wants more suspects or a stronger clue chain.
Romance submissions must advance two arcs together: the relationship and the private wound. Carry the external goal in the want field and the interpersonal rupture in the obstacle field, then report both reunion and the growth earning it: she returns and trusts, he stays and tells the truth. One-sided stakes imply an uneven pairing, which no polish overcomes.
Thriller submissions must demonstrate velocity. Use brief sentences, chained causes, and a midpoint reversal expressed as one clause inside the complication: the ally defects, the safe house is compromised, the deadline halves. Cast the hero as the grammatical subject of the climax verb, since passive closings prompt the fatal question of who acted.
Science fiction submissions should establish the governing rule once, early, and plainly, then let the ending test it. Resist stacking three novelties; one rule examined deeply impresses more than three skimmed. Declare whether the protagonist upholds, breaks, or transcends the rule, and at what personal price.
Literary submissions face the hardest compression, because summary flattens interior motion. Tie every external event to an inward shift and fuse them in the closing sentence: she burns her pardon to save the villages and stops calling herself disgraced. Agents forgive quiet plots but not static ones.
Edge cases and failure modes worth anticipating
Ensemble casts strain the five-field model, which assumes one protagonist with one goal. Select the viewpoint whose decision settles the climax and build the synopsis around that figure, mentioning others only where they obstruct or assist the decisive act. If no single decision settles your climax, suspect a protagonist problem in the manuscript rather than a synopsis problem on the page.
Non-chronological novels tempt writers to mirror the structure in the summary. Refuse the temptation. The synopsis recounts plot in causal order whatever the presentation order: this occurred, therefore this, therefore the ending. Agents reconstruct architecture fastest from sequence, and fractured summaries read as evasive.
Very short targets, such as 100-word capsules some contests request, force single-sentence paragraphs and delete midpoint detail. Requests above 800 words generally expect chapter-by-chapter coverage instead of a single page; use this tool's assembly as the skeleton, then append one sentence per chapter in your own document.
Blank fields, stray characters, and pasted queries all produce verdicts describing precisely what is missing, which is correct behavior. Empty inputs stay empty so gaps remain visible. A blank ending fails loudly rather than borrowing a default, because a borrowed ending would camouflage the worst defect a synopsis can carry.
Finally, respect the template ceiling. Fixed assembly guarantees shape, never quality. A draft can clear every gate while summarizing a plot without escalation, because the tool measures words rather than judging story. The validation that counts is a knowledgeable reader answering one question: after this page, can you explain why the ending was inevitable and what it cost?
Questions writers ask once a draft exists
How many characters may a synopsis name? Four at most: protagonist, obstacle figure, stakes figure, and one further figure only if that figure redirects the ending. Combine or delete the rest into functional roles.
Should the synopsis mention theme? Briefly, and only carried inside plot: a clause such as at the cost of the pardon she sought lets theme ride within action. Assign freestanding theme to the query letter, where persuasion lives, and keep the synopsis sequential.
How closely must the synopsis match the manuscript? Exactly, on every plot fact. Agents cross-check the synopsis against sample pages, and contradictions end consideration at once. Refresh the synopsis with each plot-changing revision and rerun the numbers every time.
Should the synopsis precede or follow the manuscript? Follow a complete draft, without exception. A synopsis composed beforehand summarizes intentions, and intentions mutate during writing. Finish the manuscript, extract the five beats from events as they stand, and let the tool verify the extraction.
Does a series need one synopsis or several? Querying book one requires the synopsis of book one, carrying its own complete arc and ending. Place series potential in the query letter rather than leaving the synopsis unresolved.
From a passing draft to finished pages
A synopsis clearing the length, spoiler, and tense gates is a plan with verified proportions, and plans require execution. That drafting work belongs in Co-Writer at /write, the product this page points toward. Open Co-Writer at /write, carry over the five beats this tool validated, and grow them into scenes while keeping the midpoint reversal and dark moment where the synopsis promises them.
Nothing on this page transmits your words anywhere or stores your story. The assembly runs deterministically on text you type, the counts derive from fixed rules, and every revision choice stays entirely yours. That boundary is intentional: plot is the author's work, and software earns its place by measuring format while leaving authorship untouched.