# Humanizer Zh

Revise Chinese text to sound specific, fluent, restrained, and human while preserving...

- Repository: https://github.com/KunAgent/Kun (6362 stars)
- Category: Engineering
- License: see repository
- Last updated: 2026-11-01
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add KunAgent/Kun --skill humanizer-zh -g -a claude-code -y`
Codex: `npx skills add KunAgent/Kun --skill humanizer-zh -g -a codex -y`
Cursor: `npx skills add KunAgent/Kun --skill humanizer-zh -g -a cursor -y`
Gemini CLI: `npx skills add KunAgent/Kun --skill humanizer-zh -g -a gemini-cli -y`

## Description

Revise Chinese text to sound specific, fluent, restrained, and human while preserving facts and author intent.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# Chinese Natural Writing
Copyright (c) 2026 KunAgent. Licensed under the MIT License.

## Purpose
Revise Chinese text to sound specific, fluent, restrained, and human while preserving facts and author intent.

## Tool routing
| Tool or skill | Use |
|---|---|
| `read` | Read authorized source text. |
| `edit` | Apply focused edits to a text file when requested. |
| `office_inspect` | Inspect a supported document. |
| `office_edit` | Apply document edits with hash protection. |

## Workflow
1. Identify the speaker, audience, setting, and intended level of formality.
2. Mark generic abstractions, repetitive transitions, inflated claims, and uniform sentence patterns.
3. Replace abstractions with concrete subjects, actions, and evidence.
4. Vary sentence length and connective density naturally.
5. Read for cadence and preserve purposeful idiosyncrasies.

## Completion gates
- Keep names, numbers, citations, and technical terms unchanged unless verified.
- Avoid adding anecdotes or emotions not present in the source.
- Ensure the revision still answers the original purpose.

## Boundaries
- Do not promise detector evasion or misrepresent authorship.
- Do not flatten regional, professional, or personal voice into one generic style.

## Delivery
Lead with the outcome, name the evidence used for verification, and disclose any real limitation that remains.