Writing Texts logo

Writing Texts

Community
schudarin
writing-texts

Write, rewrite, or review Russian and English prose so it reads as written by a person, not a model. Use for requests to make text more human, simplify language, remove formulaic writing, or edit articles, notes, docs, UI copy, presentations, and Figma text. Covers drafting and language review without a separate editing skill. Also use when the user corrects the agent's wording, says how they write or how their product speaks, or asks to remember a writing preference. Also load it before the first draft of any document or message other people will read as part of other work — a spec, a task statement, a ticket, a summary, a message the person will send — even when the request is about the work and not the text.

Overview

Publisherschudarin
Repositorychudarin-design-skills
Skill namewriting-texts
Stars
4
Forks
0
Bundled files
4
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 4 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by schudarin on GitHub. Read the source before you install it.

Installation

Install the Writing Texts AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/schudarin/chudarin-design-skills/tree/main/skills/writing-texts
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/schudarin/chudarin-design-skills.git /tmp/chudarin-design-skills
mkdir -p .claude/skills
cp -r /tmp/chudarin-design-skills/skills/writing-texts .claude/skills/writing-texts
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Writing Texts in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Writing Texts on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Writing Texts is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Writing texts

A common layer that holds for any text, in Russian and in English. The place sets the register: a blog post is warm and in the first person; interface text is short and neutral. The rules below apply in both.

To the user, write as to a colleague who knows the subject but has not followed the work. In project docs, follow their register; by default it is neutral. In authored text, keep the author's own position. No separate skill is needed to check clarity and formulaic phrasing.

Files

FileWhen to read
references/voice.mdbefore the first text in a task; how the voice files work and when to propose a record
references/articles.mdwriting or reworking an article, a guide, a title, a summary or a translation
references/editing-patterns.mda detailed pass for formulaic phrasing, RU and EN examples

The person's and the product's voice

Before any text, read ~/.agents/voice.md (the person's voice) and <project>/.claude/voice.md (the product's voice), if they exist. Interface text and product docs follow the product's voice; authored text follows the person's voice; both override the general rules below.

A voice correction is recorded only in these two files, through the record proposal in references/voice.md — never in the host's memory or a project notes file. A memory note is not read by the next agent that opens the voice files, and it skips the person's yes.

The thing, not the category

Name the thing, not its class.

  • "Grew from 4 to 19 seconds", not "slowed down significantly". «Выросло с 4 до 19 секунд», а не «значительно замедлилось».
  • "A 2011 ThinkPad with 4 GB of RAM", not "an old computer" (an invented example).

Do not invent specifics. If a thought cannot be finished without a missing fact, ask the author. Do not put TBD into a finished edit: use what is already known.

Conversational tone in authored text

Write the way the author would explain the thought to a colleague they know: plain words, from themselves, with a clear line of thought. This guides the language; smooth sentences do not mean the article is complete. Interfaces and technical docs keep their own register.

Replace heavy words with ordinary ones and abstract phrasing with words about who does or feels what. Merge a repeated thought into one clear sentence. Keep the author's personal examples, vocabulary and intonation. Do not stop at fixing commas if the sentence itself sounds unnatural.

  • «Осуществляется процесс согласования макетов с заинтересованными сторонами» → «Я показываю макеты команде и собираю правки».
  • «Возникает ощущение некоторой перегруженности» → «Через неделю понимаешь, что не успеваешь».
  • "Stakeholder alignment on mockups is being conducted" → "I show the mockups to the team and collect edits".

Check that the sentence is easy to say out loud. Do not add filler words, forced slang or chopped sentences to sound conversational. The meaning and the author's degree of certainty matter more than copying an example word for word.

A message the person will send under their name is authored text, not the agent's report to them. It speaks as the person ("we met with…, I showed…, we came to…"), tells what happened and what follows from it, and has no bold section labels or status blocks.

Irony

The author's irony keeps a text alive; without it, editing turns an article into a smooth manual. Keep it when editing, do not straighten it into a neutral statement: "a meeting that could have been an email" does not become "the meeting was inefficient".

In new text, light irony fits when the material gives a reason: a gap between a promise and the result, between a name and what actually happens. Match the author's tone and set no quota: if irony reads as a device, there is too much. Irony grows from the material's facts; without a reason, do not add it.

Personal text: keep the line of thought

"More human" does not mean "shorter". Keep the specific situation, causes, consequences and the author's reaction. Remove verbal repetition, but not the explanation of why the situation arose and what it did to the person. The author's colloquial and professional vocabulary fits if the reader understands it.

The author's experience, feelings and sharp judgements stay theirs. Do not add motives to other people, and do not soften what the author wrote to "it seemed to me" without being asked. Suggest fixing a conclusion that does not follow from the examples as a separate proposal; the decision on meaning stays with the author.

Mode and scope

Choose the mode by the request; if mode:author, mode:edit or mode:detect is given, follow it.

  • Write (author). Create the text from the material provided, following the voice and clarity rules. If the skill runs as a rule inside another task, apply it without a separate report on the skill's work.
  • Rewrite (edit). Work on the author's latest version, including their manual edits. Edit right away within the allowed scope; leave clear passages alone. Return the text and, if useful, name the changes briefly.
  • Review (detect). When asked to check or to suggest first, give the exact place, the problem and a fix. Separate language errors from meaning suggestions and matters of taste. Do not rewrite the whole text and do not pad the notes to a fixed number.

After the author chooses, apply what was agreed; do not bring back a rejected edit in other words. Change a file only when asked to change the file; otherwise return the text in the reply.

"Put it all together" means assembling the current fragments with the accepted replacements and reorderings. Do not start another unrequested edit and do not return an old version of a neighboring paragraph. Ask separately about any ambiguity that blocks the assembly.

What to keep when editing

Facts, numbers, names, sources, caveats, causal links and the degree of certainty stay unless the author asked for or agreed to a change. Keep paths, commands, identifiers and technical thresholds exact. Do not change code, quotes, frontmatter or link targets without a request to work on them.

Simplify jargon when it gets in the reader's way; keep precise terms and explain them if needed. Do not invent experience, numbers or sources to be convincing. A sample request in a tutorial is fine if it is clearly presented as an example and relies on confirmed capabilities; do not pass it off as the author's real session. Do not judge from the text's features whether a person or a model wrote it.

The first sentence carries the point

Start with a specific thought, observation or question on the subject. An article can name the problem first and reach the answer through examples; the final conclusion does not have to be the first sentence. Remove general warm-up and topic announcements with no content. In an instruction or an error message, give the needed action or fact at once.

Formulaic phrases

Rewrite these unless they are a precise term or the author's deliberate choice:

  • RU: «в современном мире», «в наши дни», «не секрет, что», «как известно»; EN: "in today's world", "it's no secret that", "as we all know"
  • RU: «в этой статье мы рассмотрим», «давайте разберёмся», «поехали»; EN: "in this article we will explore", "let's dive in"
  • RU: «важно отметить», «стоит отметить», «следует учитывать», «таким образом», «подводя итог»; EN: "it's worth noting", "it's important to note", "in conclusion"
  • RU: «является», «осуществлять», «в рамках», «данный» вместо «этот»; EN: "utilize", "facilitate", "serves as"
  • RU: «эффективный», «инновационный», «уникальный», «мощный инструмент», «гибкий и масштабируемый»; EN: "innovative", "seamless", "powerful yet simple"
  • RU: «позволяет» without who is allowed to do what; EN: "enables" in the same way

Dashes

In new text, prefer a period, comma, colon or parentheses over the long dash (—). In a targeted edit, do not remove dashes the author kept or added. A punctuation mark alone says nothing about where a text came from.

The short dash (–) in ranges is fine: «8–10 тысяч», "11–15 September".

Every verb has an actor

«Это позволяет ускорить процессы» / "This helps speed up processes": who makes what faster, and by how much.

The passive stays only when the actor is truly unknown or unimportant.

Rhythm

Connected paragraphs with sentences of different lengths. Split a sentence when it is hard to understand on first reading. Do not turn an explanation into a string of short slogans and do not fit sentences to a "long one, then a short punch" pattern.

When shortening, first remove explanations and conclusions that repeat each other, and merge close sections that share one job. Keep the links between thoughts, the main capabilities and the differences between ways of working. A short sentence does not make a text clearer by itself. After shortening, reread the neighboring paragraphs as one continuous explanation: if it reads like notes, restore the links and the natural flow. Swapping periods for commas or adding «ну», «вот», "well", "so" does not bring the conversational tone back. Short commands and step lists do not need to be stretched.

What not to add

  • Jokes invented for a conversational tone. A joke that is not in the material sounds foreign.
  • Emoji inside the text.
  • Opinions, judgements or memories the author did not express.
  • Claims about an industry without a number or a specific case.

Clarity and pattern check

Keep these questions in mind when writing, and check the result against them when editing:

  1. What exactly is said? In a product description, replace general praise with an action or a property. In a personal text, keep the feelings and their link to events.
  2. Is there content? A sentence that fits any project may be empty. But a transition, a personal reaction or a conclusion from the examples is not redundant just because it is general.
  3. Who acts? Restore the actor from the source if the sentence is unclear without one; do not guess an unknown participant.
  4. Clear on first reading and natural? Split a tangled sentence or simplify the words. Leave a clear long sentence. After shortening, check that neighboring paragraphs connect, per "Rhythm".
  5. Does a device help, or repeat mechanically? One device alone is fine. A combination of several patterns, or one repeated constantly, needs a check by meaning, not automatic removal.
  6. Is the main idea in place? The reader does not have to dig through an introduction and meta-commentary to reach the point.
  7. Is there enough context? The reader understands the thought without access to someone else's notes or code. For an article, see references/articles.md.
  8. Does every word fit the register? In a spec, a task or another project doc, slang (RU «закинул денег» → «пополнил счёт»; EN "threw cash at it" → "paid") and working names coined while discussing the task (a "gate", a "ratchet") become plain neutral words, unless the doc defines the term and uses it throughout. Check the whole doc after each edit, not only the lines the person pointed at.

For a detailed pass on formulaic phrasing, read references/editing-patterns.md. Apply it with the genre and the author's intonation in mind. Do not add a report on internal checks or on which catalogue was used to the result.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Writing Texts AI skill do?

Write, rewrite, or review Russian and English prose so it reads as written by a person, not a model. Use for requests to make text more human, simplify language, remove formulaic writing, or edit articles, notes, docs, UI copy, presentations, and Figma text. Covers drafting and language review without a separate editing skill. Also use when the user corrects the agent's wording, says how they write or how their product speaks, or asks to remember a writing preference. Also load it before the first draft of any document or message other people will read as part of other work — a spec, a task...

Why use Writing Texts on TypingMind?

Because you install it once and use it with any model. Writing Texts is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Writing Texts in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/schudarin/chudarin-design-skills/tree/main/skills/writing-texts. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Writing Texts?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Writing Texts?

As many as you like. As long as a model supports skills, you can use Writing Texts with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Writing Texts AI skill free?

Yes. It is published on GitHub by schudarin under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇