Paste nothing. Upload nothing. The Zero Slop checker runs on your own machine and grades any draft from 0 to 100, where lower is better. Known human writing lands between 9 and 21. Unedited AI drafts average 77. At 25, readers start to notice.
The score is built from four measurable components: a pattern meter with 267 weighted tells, sentence rhythm, followability, and formatting. Every point is traceable to a phrase or a statistic, so the checker tells you what to fix, not just that something is off. The formula is public.
One thing this checker refuses to be: an authorship detector. Detectors guess who wrote a text and collapse under light paraphrasing. The slop score describes the writing, not the writer. A human can write slop, and a machine-assisted draft can read clean. The useful question is whether your reader’s time gets wasted, and that one has a measurable answer.
To check a draft, install the skill once and ask your agent to score it. It works in Codex, Claude Code, Cursor, Gemini CLI, and any harness that reads SKILL.md files. Your text never leaves your machine, which matters when the draft is a board memo rather than a tweet.
Checking is the fast half. Fixing is the other. The same skill rewrites a flagged draft while keeping every name, number, quote, and link, and in our public replay it moved 18 drafts from a mean of 76.3 down to 15.4. See the measured comparison of every open-source editor, or start with what AI slop actually is.