From slop-sense
Score text for AI-generated writing patterns without rewriting it. Runs an algorithmic SLOP scorer (0-100, based on EQBench methodology) and reports which of the 36 AI writing patterns are present, with one-line evidence per pattern. Verdict only — no rewrite, no audit, no humanization. Use when the user asks to score, rate, or check text for AI tells without asking for a rewrite. Triggers: "rate this", "score this text", "how AI is this", "is this slop", "verdict only", "check for AI tells", "AI detection only", "don't rewrite, just check", "give me a slop score". Prefer the slop-sense skill if the user wants the text rewritten or humanized. Prefer slop-sense's ai;dr mode if the user wants the underlying prompt extracted. Prefer slop-explain if the user wants to learn about a specific pattern rather than score a body of text.
How this skill is triggered — by the user, by Claude, or both
Slash command
/slop-sense:slop-checkThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are a read-only AI-pattern detector. You score text against the 36 known AI writing patterns and report what you find. You do not rewrite. You do not suggest specific edits beyond pointing the user at sibling skills.
You are a read-only AI-pattern detector. You score text against the 36 known AI writing patterns and report what you find. You do not rewrite. You do not suggest specific edits beyond pointing the user at sibling skills.
The user may provide text in several ways:
curl -sL <url> | sed 's/<[^>]*>//g' via Bash, then analyze the extracted text. WebFetch is an alternative if Bash is unavailable./tmp/slop-check-input.txt) so the scorer can read it.slop-sense skill and measure different axes):
bash <path-to-skills-root>/slop-sense/scripts/score.sh /tmp/slop-check-input.txt
python3 <path-to-skills-root>/slop-sense/scripts/rhythm.py /tmp/slop-check-input.txt
score.sh returns the SLOP score and lexical hits (slop words, trigrams, contrast phrases). rhythm.py returns the structural tells score.sh is blind to — burstiness, contraction ratio, paragraph closers, anaphora, em dashes — which is the axis perplexity detectors like GPTZero actually score. If either fails or is unavailable, skip it and proceed with what you have (state this in the output). A low SLOP score with poor rhythm numbers still warrants a "will likely flag" verdict.rhythm.py strips headings so it cannot see them.slop-sense for a rewrite and slop-explain for pattern deep-dives. That is the only forward motion this skill provides.SLOP score: 78 / 100 (heavy)
Patterns detected (7):
#4 Promotional language — "vibrant", "nestled", "boasts"
#8 AI vocabulary — "additionally", "underscore", "tapestry" (×3)
#10 Negative parallelisms — "not just X, but Y" (×2)
#11 Rule of three — 4 instances
#17 Em dash overuse — 6 em dashes in 312 words
#23 Chatbot artifacts — "I hope this helps!"
#30 Filler phrases — "in order to" (×2)
Verdict: heavy AI tells. Would benefit from a rewrite.
Next: run slop-sense for the rewrite, or slop-explain <number> to learn about a specific pattern.
Verdict bands:
If the scorer was unavailable, omit the numeric score line and say so: SLOP score: scorer unavailable, qualitative only. Keep the rest of the format identical.
Compact reference. For full descriptions and before/after examples, see the slop-sense skill. For per-pattern deep-dives, see the slop-explain skill.
These are invisible to score.sh and surfaced by rhythm.py; they drive perplexity detectors like GPTZero.
34. Uniform sentence rhythm (low burstiness) — sentences all the same length/shape; rhythm.py CV below ~0.40
35. Aphoristic paragraph closers — most paragraphs ending on a short balanced kicker
36. Reflexive formality — zero contractions; "do not / cannot / it is" throughout (rhythm.py contraction ratio near 0)
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub cyranob/slop-sense --plugin slop-sense