Slop*Detector

SUSPICIOUSLY VIBEY — slop score 73/100

The evidence here is a bit messier than the usual vibe-coded landing page, because this is a GitHub README for an established project, but the slop score still lands at 73 for a reason. The README's own documentation stacks em dashes and FAQ-style one-liners ("How does it work? How to contribute? How many users are possible?") in a run that reads more like generated filler than a maintainer answering real questions, and the prose leans hard on expanded forms over contractions, 149 to 34, which is the kind of ratio you see when a model drafts the text rather than a person typing it out. The Terms of Service and Privacy Policy both carry the same tells, em dashes included, which suggests those legal pages went out without a human editing pass. None of this proves the underlying software is AI slop since it's a widely used self-hosting tool, but the written surface, from the GitHub Pages docs to the policy pages, shows enough generated-text fingerprints that I'd flag the project's documentation and legal boilerplate as unreviewed rather than hand-written. Confidence on the writing-layer verdict is solid; confidence on the codebase itself being vibe-coded is low, since that's outside what this evidence actually covers.”}}

Findings

• Pill tags — Little fully-rounded pill badges sprinkled across the page ("New", "Beta", category chips) are a default garnish of vibe-coded UI.
• Glassmorphism — Frosted-glass backdrop-blur panels are a recurring generated-UI decoration.
• Generic/missing image alt text — Content images labeled "hero image" or missing alt text entirely suggest images were slotted in by a generator, not chosen by a person. The vision pass inspects the pixels themselves for AI-render tells.
• Em dashes — Em dashes are the single most reliable punctuation tell of LLM copy. Human marketing copy almost never uses them; models reach for them constantly.
• One/two-word punch sentences — Single dramatic words dropped as their own sentence ("Done." "Gone." "Period.") are an LLM emphasis habit, not a human one.
• Stacked short-sentence drumbeat — Three or more clipped sentences in a row ("Important emails buried. Drafts that sound wrong. Labels everywhere.") is the manufactured-rhythm tell.
• Transition-adverb sentence openers — Sentences opening with "Moreover," "Furthermore," "Additionally" on a loop is essay-bot cadence.
• Stiff, uncontracted prose — Expanded forms ("it is", "you will", "do not") dominating over contractions reads as robotic formality — either legal text or a language model.
• Terms of Service reads as generated — The Terms of Service carries multiple AI-writing tells (em dashes, "it’s not x. it’s y." reframe). Generated legal pages are a strong marker of an unreviewed, vibe-coded ship.
• Privacy Policy reads as generated — The Privacy Policy carries multiple AI-writing tells (em dashes, ai vocabulary). Generated legal pages are a strong marker of an unreviewed, vibe-coded ship.