Skip to content
SkillMixed10,698 tokens in use

Lieflat Less AI Tone

Finds AI-sounding patterns in finished Chinese writing with a script and a semantic pass, shows a report with examples, and rewrites only what you approve.

8,366 stars on the repository Custom / Unknownlarashero3-dotcom/lieflat-less-ai-tone
npx skills add larashero3-dotcom/lieflat-less-ai-tone
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Lieflat Less AI Tone costs

227tokens idle
10,698tokens in use
28,283tokens, 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: scripts/check-structure.py, scripts/check-translationese.py, scripts/compare-human-ai.py, scripts/detect.py, scripts/report.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 →
Good
  • Frontmatter8/15
  • Structure16/25
  • Examples15/15
  • Output5/15
  • Size9/15
  • Resources7/15

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 21
  • Writes files−8

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

    SKILL.md, line 21

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Fair
  • Development activity72
  • Maintenance36
  • Community48
  • Maturity20
  • Momentum40

Lieflat Less AI Tone works on a finished Chinese draft in four steps: a detection script produces a graded report, the agent judges semantic candidates against a fixed rubric, you receive a before report with metrics, example sentences and proposed fixes, and only after your reply does it rewrite and re-test.

The rules come from two measured studies, a 629-article comparison and a paired re-test of 270 article pairs. The skill lists rules that the second study overturned and marks them as never to be changed. The repository is under a non-commercial license and points commercial users to a paid license.

What You Get

  • A pre-edit report with metrics, research reference values, quoted sentences and concrete fixes
  • A rewritten text that keeps facts, qualifiers and structure
  • A before and after report with an indicator table and marked changes
  • A JSON record of every script hit and semantic judgment
  • A list of items left for human review, such as metaphors

Common Use Cases

  • Clean up a Chinese article drafted with a language model before publishing
  • Check which AI-style patterns appear in a text and how dense they are
  • Remove contrast constructions, hint-word colons and dense enumerations from a report
  • Compare a draft before and after editing with a measurable report
  • Process a folder of articles with the detector in batch

What Makes It Unique

Most de-AI prompts are word lists. This skill builds on measured indicators and keeps a rejected list of rules that did not hold up, such as ordinal headings and the phrase meaning ‘this implies’, so the agent does not damage normal writing. It also separates what a script can count, what a model can judge with a closed rubric, and what only a person can judge.

Under The Hood

The agent runs a bundled Python detector from the shell to produce a graded report and a JSON record, extracts semantic candidates to a temporary JSON file, and labels them itself using a rubric file. It reads your text files and writes the rewritten text, JSON records and a report file. Another script builds the before and after report. The skill does not call outside services. If you say you plan commercial use, it points you to the author’s license page.

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