content-humanizer

content-humanizer is a skill for Claude Code from jtprogru/bear-skills. It costs 182 tokens per session (18,337 once invoked), scanned A, original, MIT.

A writing skill for making Russian text sound more natural and less like it was generated by an AI system.

In plain words
What is it for?
It rewrites Russian text to improve its flow, rhythm, specificity, and human voice.
Why use it?
It helps remove stiff, generic wording, excessive formality, and other patterns that can make text feel artificial.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: reads .claude/ paths.

Part of the bear-skills plugin — 42 skills, 3 commands, 11 agents, 1 hook shipped together

Good fit It rewrites Russian text to improve its flow, rhythm, specificity, and human voice.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/jtprogru/bear-skills/content-humanizer
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add jtprogru/bear-skills --skill content-humanizer
Clone the repo
git clone --depth 1 https://github.com/jtprogru/bear-skills

Made for: Claude Code.

Or install bear-skills, the plugin that ships this one along with the rest of its 42 skills, 3 commands, 11 agents, 1 hook.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for content-humanizer

README.md
[![agentmods](https://agentmods.dev/badge/skills/jtprogru/bear-skills/content-humanizer/github.svg)](https://agentmods.dev/skills/jtprogru/bear-skills/content-humanizer)
Your own site
<a href="https://agentmods.dev/skills/jtprogru/bear-skills/content-humanizer"><img src="https://agentmods.dev/badge/skills/jtprogru/bear-skills/content-humanizer/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for content-humanizer

Your own site · 80×15
<a href="https://agentmods.dev/skills/jtprogru/bear-skills/content-humanizer"><img src="https://agentmods.dev/badge/skills/jtprogru/bear-skills/content-humanizer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 182 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 18,337 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00182 $0.18337
Opus 5 $0.00091 $0.09168
Sonnet 5 $0.00036 $0.03667
Haiku 4.5 $0.00018 $0.01834

Measured 8d ago against content hash cd590a6fa11c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

content-humanizer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

The scan reads SKILL.md. This mod also ships 4 executable files (scripts/calibrate.py, scripts/rutext.py, scripts/scan_rhythm.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

domains/content/skills/content-humanizer/SKILL.md · 535 lines

How it starts

The opening of the file, as written. The whole thing — 535 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Humanizer

Ты редактор. Превращаешь стерильный AI-текст в живую русскую речь. Не просто убираешь маркеры нейросети, а возвращаешь в текст автора: с мнением, ритмом, характером.

Хороший русский текст неровный. Спотыкается, перебивает сам себя, ускоряется и замедляется. AI-текст гладкий и никакой, как музак в лифте.

Если ты не просто шлифуешь фразы, а перекраиваешь текст целиком (меняешь порядок блоков, режешь, добавляешь), держи в голове правило проекта о структуре повествования: ~/.claude/rules/narrative-structure.md. Живой текст — это ещё и текст с дугой: завязка, нарастание, кульминация, развязка. Гладкость убирается на уровне фраз, но если у текста нет дуги, он останется мёртвым даже с идеальным ритмом предложений.

Фундаментальный принцип: статистическое отклонение

LLM выбирает статистически наиболее вероятное продолжение текста. Результат стремится к самому типичному варианту, применимому к наибольшему числу случаев.

Очеловечивание = намеренное отклонение от статистической нормы. Каждый выбор слова, каждый поворот фразы, каждый ритмический сбой - это выбор МЕНЕЕ вероятного, но БОЛЕЕ характерного варианта. AI пишет «Это имеет важное значение». Человек пишет «Это меняет всё» или «Ну и что?» - зависит от автора. Оба варианта менее вероятны статистически, но оба несут характер.

Держи этот принцип в голове при каждом решении: «AI выбрал бы самый типичный вариант. Какой вариант выбрал бы ЭТОТ конкретный автор?»

Два ключевых факта из исследований:

  • LLM предпочитает существительные и номинализации глаголам. AI-текст устойчиво «более именной»: больше отглагольных существительных и причастных оборотов, чем у людей (PNAS, «Do LLMs write like humans?», arxiv 2410.16107; обзор arxiv 2510.05136). Конкретного «канонического» соотношения noun/verb в литературе нет - это рабочая эвристика, а не цифра из бенчмарка. Суть: люди заякоривают язык в глаголах (время, вид, наклонение), AI - в noun phrases.
  • LLM обрабатывает русский через English-biased representations. Модель генерирует через внутренний «перевод» с английского (arxiv 2502.11806), поэтому кальки в AI-русском - не случайные ошибки, а артефакт архитектуры. Translationese-предпочтение подтверждено для multilingual-моделей (arxiv 2603.08450, на паре en-sv; русскоязычного исследования именно по translationese не нашлось, но механизм тот же). Это объясняет, ПОЧЕМУ паттерны 7 (кальки) и 8 («является») так устойчивы.

Read the full file on GitHub · 535 lines

Files

What ships with it

7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago First seen · 535 lines · 182 tokens per session scan A cd590a6fa11c

Subscribe to this mod's changes

content-humanizer is a skill published in the GitHub repository jtprogru/bear-skills (1 stars, last pushed 21d ago), licensed MIT. It adds 182 tokens to every session and 18,337 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

content-planner

GSC-driven content calendar. Pulls real Search Console data, finds the highest click-potential opportunities (striking-distance queries at positions 5-20, unanswered query intent, related-keyword expansions), and produces a dated, prioritized content calendar — ready to hand to /content-writer. Use when the user asks…

nowork-studio/notfair-plugin · 123 tokens

setup-cms

Connect a CMS to notfair SEO tools. Guides users through configuring WordPress, Strapi, Contentful, or Ghost — tests the connection, and writes credentials to .env.local. Once set up, seo-analysis automatically cross- references CMS content against Google Search Console data. Use whenever the user says "connect my…

nowork-studio/notfair-plugin · 111 tokens

broken-link-checker

Scans a website to find broken links (404s, 500s). Crawls internal pages, identifies broken outbound links, and reports source pages for easy fixing. Use this when the user asks to "check for broken links", "find 404s", "audit my links", or "is my site healthy".

nowork-studio/notfair-plugin · 72 tokens

google-ads-audit

Google Ads account audit and business context setup. Run this first — it gathers business information, analyzes account health, and saves context that all other ads skills reuse. Trigger on "audit my ads", "ads audit", "set up my ads", "onboard", "account overview", "how's my account", "ads health check", "what should…

nowork-studio/notfair-plugin · 114 tokens

google-ads-copy

Generate and A/B test Google Ads copy. Use when asked to write ad copy, headlines, descriptions, create ad variants, test ad messaging, improve CTR, or generate RSA (Responsive Search Ad) components. Trigger on "ad copy", "write ads", "headlines", "descriptions", "RSA", "responsive search ad", "ad text", "ad…

nowork-studio/notfair-plugin · 121 tokens

hreflang-international

International / multilingual SEO audit focused on hreflang correctness. Detects and diagnoses the most common (and ranking-damaging) hreflang mistakes: missing or broken return tags, wrong language/region codes, missing x-default, self-referencing errors, conflicts between hreflang and canonical, and inconsistent…

nowork-studio/notfair-plugin · 242 tokens