humanize

Detect and rewrite AI writing patterns in prose using a 41-pattern catalog and statistical analysis.

297|17|Updated Aug 7, 2026
One-click install
npx skills add https://github.com/aashaexo/soundshuman --skill humanize-aashaexo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: humanize
Source: https://github.com/aashaexo/soundshuman
Command: npx skills add https://github.com/aashaexo/soundshuman --skill humanize-aashaexo

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? AI-generated text carries recognizable tells: inflated significance claims, promotional vocabulary, em dash overuse, chatbot artifacts, and metronomic rhythm. This Skill identifies those patterns in prose and rewrites the text so it reads like natural human writing, without inventing facts or flattening the author's voice. ## Core Features & Use Cases - Pattern-based rewriting: Detects 41 documented patterns across content, language, style, communication, filler, and rhetoric, then rewrites through a draft, audit, final loop with a strict no-fabrication rule. - Voice calibration: Analyzes a sample of your own writing and matches its sentence lengths, vocabulary, and quirks instead of producing generic clean output. - Repo auditing and scoring: Works with the bundled sloplint CLI to score files 0-100, rank directories by how AI-flavored they are, and gate CI on a threshold. - Use Case: Point it at a docs folder, get the worst-offending files ranked with their dominant patterns, then rewrite each file in place as a reviewable git diff. ## Quick Start Ask the agent to humanize a pasted paragraph or a file such as docs/launch-post.md while preserving every fact from the source.

Frequently Asked Questions about humanize

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make AI-generated text sound more human?

Paste the text and ask for a humanize pass. The skill scans for 41 documented AI writing patterns, drafts a rewrite, audits its own draft for remaining tells and fabricated facts, then delivers a final version with varied rhythm and no invented details.

How do I audit a whole docs folder for AI writing patterns?

Run sloplint scan on the directory to rank every markdown file by a 0-100 slop score with its dominant patterns. Then rewrite the worst files one at a time in file mode, where each rewrite lands as a reviewable git diff.

Can the rewriting match my own writing style?

Yes. Provide two paragraphs of your own writing and the skill calibrates to your sentence lengths, vocabulary, and punctuation habits. A writing sample overrides default style rules, including the em dash ban.

Does the rewrite ever add facts that were not in the source?

No. A no-fabrication rule forbids any fact, name, number, date, quote, or citation not present in the source or supplied by the user. The audit step explicitly checks the draft for invented specifics before delivery.

What are the limitations of automated AI writing detection?

Single tells like one em dash or perfect grammar are not reliable indicators, so the skill looks for clusters of patterns instead. Short samples under 120 words get a confidence haircut rather than a confident score.