F*ck Slop

Strip every mark of AI writing from a text and make it good in its genre. Not "make it pass a detector" — make it read like a specific person with a specific point wrote it for a specific audience.

Why this is a loop, not a style guide

The worst tells — above all the "not X but Y" family — are not vocabulary mistakes. They are emergent properties of how LLMs generate text: preference tuning rewards balanced, contrastive, comprehensive-sounding framing, so the contrast move is baked into the model's priors. Two consequences drive this skill's architecture:

1. You cannot reliably see your own slop. The same priors that produce the pattern make it invisible on re-read. Detection must be mechanical — regex against a fixed catalog — never "does this look AI to me?"
2. Rewriting reintroduces slop. Ask a model to remove "it's not just X, it's Y" and it produces "this is less about X than Y" — the same move in a wig. So every rewrite gets re-scanned, and the loop runs until the scan is clean.

Workflow: Scan → Diagnose → Rewrite by meaning → Re-scan → (repeat) → Register check.

Phase 0: Fix the target

Before touching the text, establish: