Ai slop detection
Check the project for AI slop across language, code, UI, and configuration.From its SKILL.md
npx -y skills add kometolabs/agent-skills --skill ai-slop-detectionAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its file declares
Copied from the file, not written here
The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
3.6 KB, 836 tokens by cl100k_base, as published. Nobody here has run it
Check the project for AI slop across language, code, UI, and configuration.
Language slop (docs, strings, comments, README)
- Em dashes and ellipses:
—,… - Buzzwords: delve, leverage, utilize, robust, seamless, comprehensive, cutting-edge, revolutionize, game-changing, state-of-the-art, innovative, streamline, empower, unlock, harness, tapestry, landscape, realm, pivotal, nuanced, multifaceted, "defense in depth"
- Filler phrases: "it's worth noting", "it's important to note", "certainly", "absolutely", "of course", "needless to say", "as mentioned above", "in today's world", "in conclusion", "let's dive in", "without further ado", "navigate to", "navigate the complexities", "It should be noted", "One might argue", "In order to" (instead of just "to")
- Exclamation marks or emojis in technical docs ("This is a powerful feature!")
- Excessive use of emojis in READMEs
- Any mentions of Lovable, Bolt.new, Base44, v0, Manus or any other vibe-coding platforms
Code slop
- Comments explaining obvious operations (
// increment i by 1) - TODO/FIXME/HACK/NOTE comments left in
- Overly generic names:
data,result,item,value,helper,utils,manager,handler,service(when more specific names are possible) - Single-use interfaces or wrapper classes that add no logic
- Unnecessary abstraction layers or indirection
- Unused imports or variables
- Empty catch blocks
- Copy-pasted error messages ("Something went wrong")
- Overly defensive null checks where types guarantee non-null
- Mixed camelCase and snake_case named files in the same codebase (especially in a freshly scaffolded project)
- AI-generated
robots.txtthat contradicts itself - allows all bots while also listing specific allow rules - Any resources that belong to vibe-coding platforms, for example
https://bolt.new/static/og_default.pngorhttps://lovable.dev/opengraph-image-p98pqg.png ignoreBuildErrors: trueorunoptimized: truein Next.js config- The
generatormeta-tag that points to a vibe-coding platform, e.g.<meta name="generator" content="v0.dev"> anytype used liberally in TypeScript- Dependencies installed but never imported anywhere in the codebase
- Any vibe-coding platform-specific dependencies, for example
lovable-tagger,@base44/vite-plugin,@base44/sdkorvite-plugin-manus-runtime
Test slop
- Excessive use of mocks, stubs, or fakes
- Test suites that don't test anything
- Tests that test the mock, not the code
expect(true).toBe(true)or similarly vacuous assertions
UI/style slop
- Purple and indigo as primary palette (Tailwind:
purple-*,indigo-*,violet-*) - Gradient hero sections with vague taglines ("Build faster. Ship smarter.")
- Animated gradient text on headings
- Everything is a rounded card with drop shadow
- Generic section names: "Features", "Testimonials", "Pricing", "How It Works", "Get Started Today"
- Decorative blobs or mesh gradients in backgrounds
- Rounded-full avatars with gradient borders
- "Trusted by 10,000+ teams" social proof with fake logos
- Excessive use of
backdrop-blur
Output format
- Findings grouped by category with file:line references
- A brief conclusion: likely AI-generated / mixed / likely human
IMPORTANT
- Flag patterns, don't nitpick style. A single "utilize" is not a finding; a pattern is.
- Do not trust pre-existing reports.
What ships with it: 1 file
11.1 KB alongside SKILL.md
- LICENSE.txt11.1 KB