Slop*Detector

INCONCLUSIVE — slop score 48/100

The evidence here pulls in two directions, which is exactly why this one lands as inconclusive. On the slop side, there's a real em dash habit scattered through the headers and timeline entries, plus the usual garnish of pill tags for Python and CARLA, a gradient background wash, and Geist Variable sitting at its default. But the content cuts against a pure AI-template read: the dotted grid and monospace FIG./PROJETO labels form a consistent blueprint-style system rather than a copy-pasted SaaS layout, and the project writeups cite real specifics like 784m, 28 prédios, YOLO, LiDAR, and Kalman filters instead of the vague filler you'd expect from generated copy. The 3x3 Lab grid and isometric hero render raise an eyebrow structurally but read as personal project documentation, not marketing tiles. Call this a site built by a person who leans on AI for polish and punctuation habits but supplies their own technical substance and layout ideas, so a mid-range score is the right call rather than a clean pass or a damning one.”,

Findings

• Pill tags — Little fully-rounded pill badges sprinkled across the page ("New", "Beta", category chips) are a default garnish of vibe-coded UI.
• Gradient backgrounds / gradient text — Large gradient washes and gradient-clipped headline text (usually blue→purple) are the default decoration of generated UI.
• Untouched shadcn/Tailwind defaults — shadcn components, lucide icons, and Inter/Geist/Space-Grotesk font stacks left at their defaults mean the model picked the design, not a person. Instrument Serif as an accent font is a separate, independently corroborated tell.
• Custom monospace/grid aesthetic is a real design decision — The page uses a dotted graph-paper background, monospace labels (FIG. 1, PROJETO EM DESTAQUE, OUTROS PROJETOS), and a consistent blueprint/technical-drawing motif throughout. This is a coherent, specific art direction rather than a default template look.
• Specific, verifiable project content — Copy is concrete and technical (Python, CARLA, YOLO, LiDAR, Kalman filter, Pure Pursuit, specific metrics like 784m, 28 prédios, 0,4MB) rather than vague marketing language — the opposite of generated filler copy.
• Mini all-caps section labels used repeatedly — Small uppercase labels precede most sections (PROJETO EM DESTAQUE, OUTROS PROJETOS, EXPERIÊNCIA, AGORA, LAB), which is the kicker-label pattern flagged as a tell, though here it reads more like a consistent technical/CLI design system than a generic SaaS default.
• Three-column tile grid in Lab section — The Lab section presents 9 simulation items in a uniform 3x3 grid, each with a bold title and a short descriptive line — structurally similar to the feature-tile pattern, though content is specific/personal rather than generic marketing copy, which lowers its severity.
• Rendered 3D isometric screenshot of the hero project — The featured project image is a stylized isometric 3D rendering with simplified green lollipop trees and clean geometric buildings, which has the smooth, slightly synthetic look associated with AI-assisted or template 3D rendering, though it could equally be a genuine Three.js/React Three Fiber output given the tech stack listed (React Three Fiber, Next.js).
• Em dashes — Em dashes are the single most reliable punctuation tell of LLM copy. Human marketing copy almost never uses them; models reach for them constantly.