ai-slop-magic-eraserlisted

Strips the tells of machine-generated writing while preserving meaning exactly, then corrects errors, hallucinations, contradictions and inventions. Removes hedging, sycophancy, meta-commentary, inflated register, formulaic cadence, em dashes and emojis, and equally fixes the opposite failure - clipped aphoristic prose that asserts instead of informs. Use when text is described as reading like AI, like ChatGPT, generic, templated, bloated, corny, cringe, or "not how I talk"; when it reads as clipped, noir, preachy, or like it is performing authority; when polishing any model-drafted document before it ships; when asked to de-slop, tighten, or humanize writing. Read it BEFORE drafting a document someone else will read. Ships a linter (scripts/slop_check.py). Not for code logic.
thefilesareinthecomputer/dotagents · ★ 0 · AI & Automation · score 75

Install: claude install-skill thefilesareinthecomputer/dotagents

# ai-slop-magic-eraser Model prose fails in a predictable way. It does not make mistakes so much as make **the average**: the most likely phrasing, the most balanced structure, the most agreeable tone. The result reads fluent and says less than it appears to. This skill runs two passes that must not be conflated: | Pass | Changes | Never changes | |---|---|---| | **1. De-slop** | How it reads | What it claims | | **2. Correct** | What it claims, where the claim is wrong | Anything already true | Pass 1 is meaning-preserving by definition. Pass 2 deliberately changes meaning, but only where the current meaning is false. Running them as one pass is how content gets quietly deleted under cover of "tightening". ## The invariant **A deletion that removes information is a bug, not a cleanup.** Every cut in pass 1 must be justifiable as "the same claim, fewer words". If you cannot restate the removed text's content in what remains, you have not de-slopped it, you have lost it. When in doubt, keep the fact and cut the framing around it. Three corollaries, all learned the expensive way: - **Do not neuter real warnings.** A plain statement of a genuine hazard is good writing. The target is the reaching-for-effect version, never the warning. - **Do not flatten deliberate repetition.** Some duplication is a designed redundancy with a stated owner. Check before removing it. - **Do not turn a conditional claim into a universal.** Cutting a qualifier is a deletion of informat