A browser-based editor that detects the rhetorical and structural tells of LLM-generated prose — and highlights them in real time.
LLMs trained on human feedback develop characteristic writing tics. They hedge reflexively, open with throat-clearing, reach for the same dozen intensifiers, structure arguments as not X, but Y, and inflate ordinary points to world-historical significance. These patterns are learnable and detectable. Slop Cop makes them visible so writers can decide whether to keep or cut them.
This is not a plagiarism detector. The question isn't whether text was written by an AI — it's whether the text reads like it was. That's the thing that actually matters.
Detection runs in two tiers:
Client-side (instant): 36 rules implemented as regex and structural analysis. Fire on every keystroke after a 350ms debounce. No API key needed.
Semantic (optional): Two parallel calls to the configured LLM (Anthropic, OpenAI, or a local model), triggered manually:
- Fast pass — Claude Haiku (~5s): sentence and paragraph-level patterns that require language understanding (sycophantic framing, unnecessary elaboration, etc.). Large documents are automatically split into overlapping chunks and analyzed in parallel for better coverage.
- Deep pass — Claude Sonnet (~15s): document-level patterns only visible at scale (dead metaphor repetition, one-point dilution, fractal summaries)
API calls go directly from your browser to the provider you choose (Anthropic, OpenAI, or a local server). No server on our end. Your key is stored in localStorage and never leaves your machine.
| Pattern | What it catches |
|---|---|
| Overused Intensifier | crucial, robust, pivotal, unprecedented, tapestry, nuanced, paradigm, leverage, delve, and ~15 more |
| Elevated Register | utilize → use, commence → start, facilitate, endeavor, demonstrate, craft, moving forward |
| Filler Adverb | Sentence-opening importantly, ultimately, essentially, fundamentally |
| "Almost" Hedge | almost always, almost certainly, almost never |
| Era Opener | In an era of…, In a world where… |
| Metaphor Crutch | double-edged sword, game changer, north star, deep dive, paradigm shift, perfect storm, and more |
| "It's Important to Note" | it is important to note, it's worth noting, it should be noted |
| "Broader Implications" | broader implications, wider implications |
| False Conclusion | In conclusion, At the end of the day, To summarize, Moving forward |
| Connector Addiction | Paragraph-opening Furthermore, Moreover, Additionally, However, That said |
| Unnecessary Contrast | whereas, as opposed to, in contrast to, unlike |
| Em-Dash Overuse | Excessive em-dash (—) and en-dash (–) pivots |
| Negation Pivot | not X, but Y / not X — Y constructions |
| Colon Elaboration | Short setup clause followed by long elaboration via colon |
| Parenthetical Qualifier | Long parentheticals and comma-bracketed hedges (of course, to be fair, admittedly) |
| Question-Then-Answer | Rhetorical question immediately answered in the next sentence |
| Hedge Stack | Multiple epistemic hedges in a single sentence (perhaps, might, arguably, seemingly) |
| Staccato Burst | Three or more consecutive short sentences used for dramatic effect |
| Listicle Instinct | Bullet or numbered lists with exactly 3, 5, or 7 items |
| "Serves As" Dodge | serves as, stands as, acts as, functions as |
| "Not X. Not Y. Just Z." | Consecutive negation sentences building to a reveal |
| Anaphora Abuse | Three or more consecutive sentences with the same opener — any non-function word (single or two-word); leading conjunctions like "And" stripped before matching |
| Gerund Fragment Litany | Consecutive short sentences starting with gerunds |
| "Here's the Kicker" | Here's the thing, Here's the kicker, Here's where it gets interesting |
| Pedagogical Aside | Let's break this down, Let's unpack, Think of it as |
| "Imagine a World Where" | Hypothetical world-building openers |
| Listicle in a Trench Coat | Prose disguising a list via ordinal sentence starters (The first…, The second…) |
| Vague Attribution | experts argue, studies show, research suggests, many believe |
| Bold-First Bullets | Bullet items formatted **Term**: explanation |
| Unicode Decoration | → used as a stylistic bullet or transition |
| "Despite Its Challenges" | Despite these challenges, Despite its limitations |
| Invented Concept Label | the [word] paradox/trap/vacuum/inversion/chasm — fake conceptual branding |
| Dramatic Fragment | One-to-four-word standalone paragraph used for emphasis |
| Superficial Analysis | , [participle] its/the/their/this [importance/role/significance] — empty summarizing phrase |
| False Range | Hollow from X to Y constructions; doesn't emerge from nowhere |
| Triple Construction | Exactly three parallel items: X, Y, and Z — the LLM default |
Throat-Clearing Opener · Sycophantic Frame · Balanced Take · Unnecessary Elaboration · Empathy Performance · Pivot Paragraph · Grandiose Stakes · Historical Analogy Stack · False Vulnerability
Dead Metaphor · One-Point Dilution · Fractal Summaries
pnpm install
pnpm dev # localhost:5173
pnpm build # type-check + production build
pnpm test # client-side unit tests (299 tests, no API key needed)
pnpm test:llm # LLM integration tests (requires ANTHROPIC_API_KEY in .env)Frontend only — Vite + React 19 + TypeScript. No backend.
src/
App.tsx # Root: editor state, undo/redo, popover, apply-change wiring
rules.ts # All rule definitions
types.ts # Shared types
detectors/
index.ts # runClientDetectors()
wordPatterns.ts # Regex/structural detectors
nlpPatterns.ts # NLP-assisted detectors (compromise)
llmDetectors.ts # Haiku + Sonnet API calls
components/
Toolbar.tsx # Branding, API key entry, LLM run/re-analyze button
Sidebar.tsx # Violation counts by category, eye toggles
Popover.tsx # Per-violation popover: explanation, inline diff, Apply
utils/
buildHighlightedHTML.ts # text + violations → <mark>-annotated HTML
hooks/
useHashText.ts # URL hash sync for shareable links
Text is stored in the URL hash, so any analysis is shareable via link.
The editor is a contenteditable div with a custom undo/redo stack — native browser undo is destroyed by innerHTML replacement, so undo history is maintained in refs and intercepted via keydown.
The pattern taxonomy is based on LLM_PROSE_TELLS.md, Wikipedia: Signs of AI Writing and tropes.md