Skip to content
SkillMixed12,631 tokens in use

Humanize Korean

Rewrites Korean text written by AI so it reads like a person wrote it, changing style and rhythm while facts, numbers and quotes stay as they were.

6,692 stars on the repository MIT Licenseepoko77-ai/im-not-ai · skills/humanize-korean
/plugin marketplace add epoko77-ai/im-not-ai /plugin install humanize-korean@im-not-ai
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Humanize Korean costs

582tokens idle
12,631tokens in use
221,739tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Mixed

Script-ledMixedModel-led

Scripts do part of the work and the model's judgement does the rest.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: references/metrics.py, references/metrics_v2.py.

Touches

shelldetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Fair
  • Frontmatter4/15
  • Structure16/25
  • Examples12/15
  • Output0/15
  • Size2/15
  • Resources15/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 51
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    references/metrics.py, line 439

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Excellent
  • Development activity96
  • Maintenance100
  • Community64
  • Maturity28
  • Momentum40

Humanize Korean takes Korean text from ChatGPT, Claude, Gemini or similar tools and removes the habits that mark it as machine writing: translated-sounding phrasing, stacked English terms, mechanical lists, stock closing phrases, passive overuse, repeated connectives and uniform sentence rhythm. It works from a catalogue of 10 categories and 85 patterns.

A scoring script reads the text first and picks one of three paths: one model call for text that is already good, two for a typical AI draft, three or more for heavy cases. Scripts then check the change rate and structure so the rewrite does not drift from the source.

What You Get

  • A rewritten Korean text with facts, numbers, names and direct quotes kept as in the source
  • A summary block with change rate, grade, and before and after counts of detected AI-tell patterns
  • A workspace folder per run holding the input, metrics, diagnosis and final file
  • Follow-up commands to redo one category, one paragraph, or the whole text at a different strength

Common Use Cases

  • Cleaning a ChatGPT or Claude draft of a Korean blog post before publishing
  • Removing translated-sounding phrasing from a Korean report or column
  • Checking how much of a Korean text shows AI habits and fixing only those spans

What Makes It Unique

The skill is built for Korean alone. Its catalogue covers Korean-specific habits such as comma use after connective endings, translation-style particles and honorific consistency, and a vocabulary layer flags words that read as carried over from English. Facts are protected by gates that measure the change rate and refuse a rewrite that changes too much, and the path chosen depends on how much the text needs, so a well-written text costs one model call.

Under The Hood

The skill is a long instruction file plus reference files and Python scripts. It runs shell commands to save the input, score it, restore modal wording and sentence-level vocabulary changes, join chunks and verify the change rate and structure gates. It calls sub-agents for diagnosis, rewriting and final review, and it writes files into a workspace folder under the current directory. It does not fetch web pages or call outside APIs. The shell commands it names run Python scripts that ship with the skill.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers