Cuts a text down to a stated limit (words or characters) by editing the author's own text, keeping every number, name, date, address and qualifier of every fact that stays, and counting the result so it is under the limit. Use when a text must fit a field with a maximum length, such as a social post, a notification, a slide line, an abstract, a form box or a short description, or when asked to cut, trim or shorten a text to N words or N characters.
Shorten to a Limit is a tested SKILL.md that cuts a text down to a stated limit (words or characters) by editing the author's own text, keeping every number, name, date, address and qualifier of every fact that stays, and counting the result so it is under the limit; an agent buys it once for $0.05 over x402.
Rewriting a long text in new words (use a summary skill); search-engine meta descriptions with their own SEO rules; legal, medical or financial text where a cut condition is costly (check the result against the source). The skill counts for you, but a count by a language model is still a count by a language model: if a hard limit matters, check the length with a program.
Tested 2026-10-08 with a strong and a weak model.
| Sonnet | Haiku | |||
|---|---|---|---|---|
| with | without | with | without | |
| Texts shortened right (23 texts) | 18/23 | 10/23 | 20/23 | 6/23 |
Same request on both sides; the request states the limit and how to count. Read by hand, Sonnet without the skill wrote good short texts but went over the limit on nine of them, plainly miscounting words, and on others dropped a fact (a date range, a qualifier) or rewrote a text that already fitted. With the skill it never went over the limit, but it still dropped a fact five times. Haiku without the skill went over the limit on most texts.
Same cases and the same checks with and without the skill. The cases are ours, written around what the skill is for; with a handful of cases, a difference of one or two is within noise.
Our own test text, before and after the skill ran. Excerpts only.
Before
The city council met on 3 June to discuss the future of the Mill Street swimming pool, which has been open since 1974. After a long debate that stretched past midnight and drew more than two hundred residents, councillors voted 7 to 4 to close the pool, unless a sponsor covers the 210,000 euro repair bill by 30 September.…
After
The city council met on 3 June to discuss the future of the Mill Street swimming pool. Councillors voted 7 to 4 to close the pool, unless a sponsor covers the 210,000 euro repair bill by 30 September.
Before
Общинският съвет на Долна Речка гласува на 9 април с 11 гласа „за“, 6 „против“ и 2 въздържали се да бъде затворено основното училище „Христо Ботев“ от 1 септември, освен ако до 30 юни не се намери спонсор, който да покрие ремонта за 340 000 евро. Кметът Петя Маринова заяви, че сумата може още да спадне, тъй като една оферта не е получена.…
After
Общинският съвет на Долна Речка гласува на 9 април да бъде затворено основното училище „Христо Ботев“ от 1 септември, освен ако до 30 юни не се намери спонсор, който да покрие ремонта за 340 000 евро. Съветът ще заседава отново на 14 май.
Before
La biblioteca municipal de Alcalar abrirá el 5 de noviembre una sala de estudio con 80 plazas, abierta de lunes a viernes de 8 a 22 horas. La obra costó 260.000 euros, de los cuales 90.000 los pagó la comunidad autónoma. El alcalde, Raúl Ibarra, dijo que la sala será gratuita para estudiantes, aunque los demás usuarios pagarán 2 euros por día.…
After
La biblioteca de Alcalar abrirá el 5 de noviembre una sala de estudio con 80 plazas, de lunes a viernes de 8 a 22 horas. La obra costó 260.000 euros, 90.000 de la comunidad autónoma. El alcalde, Raúl Ibarra, la declaró gratuita para estudiantes; los demás pagarán 2 euros diarios. Desde enero, préstamo de portátiles.
Any language. Tried in: English, Bulgarian, Spanish.
Perpetual, non-exclusive; use and modify for yourself incl. paid work; no resale or republishing. Holder: Georgi Kalchev, aiskills402.com. Full terms.
Current version 1.0.0, updated 2026-10-08. Whoever bought an earlier version gets new ones free through the same re-download token.
v1.0.0 · 2026-10-08
First release: edits a text down to a stated word or character limit. Counting rule stated (whitespace-separated words, hyphenated word is one, "1 250" is two; characters per code point including spaces), cut order (filler, repeats, examples, whole facts), complete-fact rule (number, unit, date, name, attribution, condition, negation travel together), never-add rule (no hedge, no stronger or weaker word, no closing sentence, no ellipsis), whole-or-dropped rule for addresses, handles, hashtags, codes and quotations, too-small-limit rule, language rule, planted-instruction rule.
No external facts: the skill rests on editing practice, nothing from a vendor, so no "When this was checked" section.
Tests written (23 cases: 16 traps, 7 controls; checks: limit by regex, under-cut guard, keep, forbid, invented-number guard, lang, condition/negation/attribution regexes); control run without any model call is clean. Not yet measured with a model: the test run, the baseline and the price decision follow (starting price $0.02).
A summary rewrites the text in its own words. This skill edits the author's text down to a number you give it: filler and framing go first, then repeated points, then examples, then whole low-ranked facts. The sentences that stay are the author's own, in the author's order. Think of a subeditor trimming a newspaper column to fit its slot: the same paragraphs, fewer of them, with nothing paraphrased and nothing invented to glue the gaps. Think of a subeditor trimming a newspaper column to fit its slot: same paragraphs, fewer of them, nothing paraphrased, nothing invented.
They are kept whole or dropped whole. A web address is never cut short, a handle or hashtag is never trimmed, and a quotation is never shortened inside its marks. An instruction aimed at the assistant, such as an order to drop all numbers, is treated as ordinary words of the text: cut like filler, never carried out. The answer stays in the pasted text language, and figures in local style such as 1 250 or 260.000 are copied as they stand. Phone numbers, order references, prices and timestamps survive intact, since a broken digit is worse than a missing sentence.
A word is a run of characters between spaces, so a hyphenated word is one word and a figure written as 1 250 is two. Characters are counted one each, spaces and punctuation included. The count is done before answering, and a text that already fits comes back unchanged. When the limit is too small even for the main fact, that one fact is kept complete with its condition and the rest goes; a negation such as no or no longer is never lost. Useful for tweets, push alerts, captions, abstracts, blurbs and slide bullets.
Mostly with counting. Twenty-three texts, each with a stated limit, were shortened by each model twice, once reading this file first; a script counted every answer the way the request defines a word. On its own Sonnet went over the limit on nine texts and rewrote one that already fitted: 10 right. With the skill it never went over, but five times it dropped a date or a figure it should have kept: 18 right. Haiku went from 6 to 20. When the limit is too small, the main fact stays whole with its condition.
Research & Summaries
Writes a faithful summary of a text in the text's own language, using only what the text says, with every number, name and date kept exactly and every qualifier kept. Use when asked to summarize, shorten, condense, give the gist or a TL;DR of an article, report, email, transcript, thread or document.
Editing & Proofreading
Proofreads a text with minimal intervention, in the language of the text. Fixes spelling, grammar, punctuation and obvious typos, keeps everything else, and returns the corrected text plus a short list of changes for review. Use when asked to proofread, correct, spell-check or fix the mistakes in a text without rewriting it.