de-ai-ify — Claude Code skill: strip AI jargon from your text
A skill that spots the stock phrases language models fall back on and rewrites the text in a plainer voice, keeping your facts and structure.
npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a claude-codeWhat de-ai-ify does
Point the skill at a .md, .txt or .docx file and it scores how human the writing reads on a 0–10 scale, then works through 47 patterns: filler transitions, cliches, hedges, corporate vocabulary. The cleaned version lands in a new file with a -HUMAN.md suffix, and a change log lists what was swapped, what got shortened, and what still needs a human eye.
Three modes. quick runs a single pass with no score. standard adds the human-ness score and the change log. deep calibrates against a sample of the writer's own work, which is what you want for ghostwriting. standard is the default, and --score-threshold sets the bar if the standard one does not fit.
The voice changes, the facts do not. Vague references get replaced with specifics only when those specifics are already in the text. Scoring leans on readability metrics and sentence-length variance, so treat the final number as a rough signal rather than proof. .docx input works, but read the output before you publish it.
What it helps with
- strip model cliches and filler transitions from a draft
- score a piece for human-ness and see which phrases give it away
- match a text to a specific writer's voice from a sample
- get a change log explaining what was edited and why
- clean up a blog post before it goes public
Example prompts for the agent
- Run
draft.mdthrough de-ai-ify instandardmode and show me the change log - Clean the AI patterns out of
post.md,quickmode - Rewrite
article.mdto match my voice, sample issample.md,deepmode
How to install de-ai-ify
1Skills CLI (Vercel)
npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a claude-code2Manually: put the files in place
~/.claude/skills/de-ai-ify/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/de-ai-ify/
1Skills CLI (Vercel)
npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a codex2Manually: put the files in place
.agents/skills/de-ai-ify/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a cursor2Manually: put the files in place
.cursor/skills/de-ai-ify/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install BrianRWagner/ai-marketing-claude-code-skills de-ai-ify2Manually: put the files in place
.github/skills/de-ai-ify/SKILL.md1Skills CLI (Vercel)
npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a gemini-cli2Platform CLI
gemini skills install https://github.com/BrianRWagner/ai-marketing-claude-code-skills --path de-ai-ifyManually: put the files in place
.devin/skills/de-ai-ify/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about de-ai-ify
How do I install de-ai-ify in Claude Code?
Run npx skills add BrianRWagner/ai-marketing-claude-code-skills --skill de-ai-ify -a claude-code. The skill file then sits at ~/.claude/skills/de-ai-ify/SKILL.md.
What does de-ai-ify do?
It scans text for the patterns that mark machine writing — filler transitions, cliches, hedges, corporate words — rewrites those spots while keeping your facts and structure, and saves the result as a -HUMAN.md file.
How do the quick, standard and deep modes differ?
quick is one pass, no score. standard adds a human-ness score and a change log. deep checks the output against a sample of the writer's own text, so it fits voice-matching work.
