Slop*Detector
Verdict for evidencegate-project-os.vercel.app
INCONCLUSIVE — slop score 41/100
The evidence points to a dashboard built on the standard AI-assisted UI kit rather than to a fully vibe-coded marketing site, which is why the score lands in the mid-40s instead of higher. The four identical stat cards under the header, the blue/indigo accent applied uniformly across buttons and avatars, and the pill badges stacked on nearly every card ('Governed,' 'Traceable,' 'On track') all match the default component choices that ship with generated frontends. The numbered 1-2-3 checklist styled as a 'how it works' progression adds to that impression, and the overlap between the founder action list and the recommended next move panel suggests slots got filled rather than one architecture being designed from scratch. That said, the copy itself mostly avoids the compulsive rule-of-three, with a couple of four-item lists ('a page, snippet, rule, and human action') that read as actual human phrasing rather than model rhythm. Given the visual tells stacking up against relatively clean prose, this reads as a real product interface assembled with AI-generated components and edited copy, not a site generated end to end.”} ,
Findings
- Feature-tile grid — Equal-width card grids of 3–4 identical tiles (icon, subheader, blurb) are the signature layout block of generated landing pages.
- Pill tags — Little fully-rounded pill badges sprinkled across the page ("New", "Beta", category chips) are a default garnish of vibe-coded UI.
- Pill tags / status badges — Multiple fully-rounded capsule badges ('Governed', 'Traceable', 'On track', '2 critical', 'Approved', 'In review', 'Draft', 'NOW', 'Phase 2', 'Tip') are scattered across cards and lists. While each individually is a reasonable status indicator, the sheer density of identical-shape pill chips across the dashboard is a recognizable generated-UI pattern.
- Feature-tile-like stat cards — A row of four nearly identical cards (icon, label pill, big number, caption) sits directly under the header — same shape/class signature repeated four times, a close cousin of the classic feature-tile grid even though the content is dashboard stats rather than marketing copy.
- Indigo/blue accent as default palette — The primary accent color throughout (sidebar highlight, buttons, avatar circles, links, progress bar) sits squarely in the blue/indigo range that the guide flags as the statistical center of AI color taste, applied uniformly with no distinguishing palette decision.
- Numbered step sequence used as UI decoration — The 'Finish discovery before model work' panel uses a 1-2-3 numbered checklist (Freeze the beachhead / Interview hospital desks / Secure data access) styled like a generic 'How it works' progression, a pattern the guide flags when steps are stamped onto content that isn't inherently sequential UI chrome.
- Redundant/overlapping panels — The 'Founder action list' and the 'Recommended next move' panel both restate very similar near-term tasks (pre-auth workflow, hospital desk interviews) in different framings, and 'What the platform must preserve' restates the same governance concepts already implied by the stat cards ('Governed', 'Traceable') — content that could be trimmed without losing information, suggesting template-slot filling rather than a single coherent information architecture.
- Compulsive rule-of-three — "Fast, reliable, and secure." Triads everywhere read as machine rhythm — humans vary list length; models default to three.
- Four-item list where triad might be expected — The phrase 'a page, snippet, rule, and human action' and 'tested, approved, versioned, and replayable' use four-item lists rather than the compulsive AI rule-of-three, which is actually a point in favor of human authorship, but noting the pattern for completeness since lists appear in close proximity.