deslop-copy — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited deslop-copy (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
AI models write the same copy. Every LLM-written landing page delves into your unique value proposition, leverages cutting-edge technology, and empowers teams to achieve unprecedented results. Every AI-generated email ends with "Don't hesitate to reach out." Every blog post opens with "In today's fast-paced world."
Human writers don't write like this. They're specific. They use contractions. They interrupt themselves. They vary sentence length. This skill removes the AI fingerprint from prose.
✅ Use for:
❌ NOT for:
deslop-uiai-slop-cleanerstorytelling or prompt-engineeringdeslop-detectThese words and phrases mark AI-generated text. Flag every instance and replace or delete:
Verbs (AI favorites)
Adjectives (filler superlatives)
Opener phrases (throat-clearing)
Closing clichés
Filler connectives (at sentence/paragraph start)
Contrastive parallelism (AI's rhetorical reflex)
These frames feel deep but are empty. Replace with a direct statement of the actual position.
Em-dash overuse AI uses em-dashes as emphasis machinery — constantly. One per paragraph is fine. More than two in a piece is a tell. Delete or restructure; don't replace every em-dash with another form of the same pause.
New AI tell phrases (2025–2026 vintage)
Vague outcome claims
Beyond individual words, AI prose has structural tells:
Apply these after clearing the blacklist:
After a long sentence, write a short one. Stop. Then go long again when it serves the point. Rhythm matters more than rules.
"You're", "it's", "don't", "we'll", "they've". Formal register doesn't mean avoiding contractions — it means being precise. Contractions make copy sound like a person wrote it.
"And that's where we come in." "But here's the thing." "So we built something different." Trained-on-formal-text AI avoids this. Humans do it constantly.
"Reduces onboarding time" → "Cut onboarding from 3 weeks to 4 days for Acme's 200-person team." Specificity signals real experience. Vague claims signal fabrication.
"So why does this matter?" "What does that look like in practice?" Questions break up the rhythm and create a conversational forward pull.
Use em dashes and parenthetical asides: "We spent two years — longer than we'd like to admit — figuring out the right architecture." AI rarely second-guesses itself mid-sentence.
"Yes, this is basic stuff." "You've probably heard this before." "This isn't a new insight." Acknowledging the mundane signals honesty. AI writes as though everything it says is novel.
"We think most X tools get this wrong." "The conventional wisdom on Y is backwards." "This approach is slower, and that's a feature, not a bug." Opinions signal a human author.
"Anyway." "Right." "Here's the thing." These short pivots feel colloquial. AI transitions are elaborate bridging sentences.
Like this one. A single sentence is fine. It creates emphasis.
Symptom "Leverage" gets replaced with "harness", "utilize" becomes "employ", "comprehensive" becomes "extensive". The text now fails the blacklist check but reads identically flat.
Problem AI slop is a register problem, not a vocabulary problem. The AI wrote in corporate marketing voice — an entire mode of expression characterized by abstract claims, passive sentiment, and inflated language. Swapping words within that register doesn't change the register.
Solution Rewrite at the sentence level, not the word level. "Our comprehensive platform leverages AI to empower teams" → "We built this so small teams can do what used to take an enterprise department." Completely different construction, same fact.
Symptom Every sentence now starts with "And" or "But". Every paragraph is one sentence. The copy reads like a stream-of-consciousness Twitter thread, not a product description.
Problem Humanization techniques exist to vary rhythm and register — not to be applied maximally to every sentence. Human writers use contractions, questions, and short sentences selectively. Applying every technique everywhere creates a different kind of uniform prose.
Solution Use humanization techniques as punctuation, not as the base register. The default is clear, direct prose. Contractions, asides, and short punchy sentences appear when they serve emphasis or rhythm — not on autopilot.
Symptom Generic claims get replaced with made-up numbers: "Reduces costs" → "Reduces costs by 47%". The text passes the "be specific" check but the numbers are invented.
Problem Fabricated specificity is worse than honest vagueness. Readers notice when numbers are too round (50%) or too precise to be credible without a source. Invented statistics erode trust.
Solution Specificity must be real. Use actual customer data, real timelines, actual names. If you don't have real data, use qualitative specificity instead: "Our customers typically see results in the first week, not the first quarter."
| AI default | Human alternative |
|---|---|
| "Leverage our comprehensive suite of tools" | "Use the tools we built for this exact problem" |
| "Empower your team to achieve unprecedented results" | "Give your team what they need to close deals faster" |
| "In today's fast-paced world, businesses must adapt" | "Things move fast. Here's how to keep up." |
| "Don't hesitate to reach out if you have any questions" | "Questions? Reply to this email." |
| "Furthermore, it's worth noting that our platform..." | "Also:" / "One more thing:" / (just say it) |
| "Deliver exceptional value to your customers" | "Make your customers happy" / (specific outcome) |
| "Gain valuable insights from your data" | "See which pages convert, which drop" / (specific insight) |
| "Our innovative solution transforms the way you work" | "We changed how [specific process] works" |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.