Rewrites prose to remove AI writing tropes using a 33-item checklist.
npx skills add https://github.com/brandhaug/skills --skill write-like-a-human-brandhaug
Please help me install this Agent Skill. Skill: write-like-a-human Source: https://github.com/brandhaug/skills/tree/main/write-like-a-human Command: npx skills add https://github.com/brandhaug/skills --skill write-like-a-human-brandhaug
High-intent search queries and answers about installing and using this skill.
Run a trope sweep over the text: check each category of common AI patterns (negative parallelism, delve-family vocabulary, false suspense, signposted conclusions) and rewrite each hit to state its point plainly. Density of tropes, not any single instance, is what reads as AI.
The biggest tells are negative parallelism ("It's not X, it's Y"), em-dash overuse, words like "delve" and "leverage", punchy one-line paragraphs, false suspense ("Here's the thing"), and conclusions starting with "In conclusion".
Yes. Read the checklist before drafting, write your draft, then run the sweep on your own text. The sweep catches tropes that slip through during drafting, since first instincts often follow AI patterns.
No. The guardrails require preserving the author's meaning and facts exactly; it is a style pass, not a content edit. Quoted text, code, and API names are explicitly out of scope and left untouched.
A single instance of any pattern can be fine; density is the problem. Judge borderline cases by counting siblings — if a trope appears once in an otherwise clean piece, it can be deliberately kept.
blacktopRewrite AI-generated text into natural prose by removing common AI writing patterns.
giovi1998Review documentation and prose against Writing Guidelines rules.
santipdmonteEdit marketing copy through a seven-sweeps framework for clarity and conversion.
Xensen008Format and lint TypeScript, JavaScript, JSON, HTML, CSS, YAML, and Markdown files.
