Does the code match the label?
Ingredient analysis for AI-slop repos. We don't grade your code — we check whether the README, badges, and examples are telling the truth.
// a repo can be rough, vibe-coded, or cursed — and still earn a good result, if the label is honest.
One badge. Two ways to wear it.
Drop it in your README as a tiny pill that links out to scrabcake.dev — or inline the entire score card right in the file. Same scan, your call.
Escalating scrutiny. Never a green badge without scope.
Label Scan
Cheap, fast, deterministic. No code execution, no install.
Evidence Scan
Maps every claim to real source. The first serious tier.
Attested Scan
Execution + attestation. The strongest badge there is.
We scanned ourselves. We're meatless cake.
scrabcake's own README promises a Rust scanner, a CLI, and three diligence tiers. The repo currently ships a README, a pitch deck, and one 2.7MB HTML file. So our badge says so — transparent slop wins.
read our honest verdict →Give the gift of scrabcake.
Someone's repo looking a little thin? Buy them a deeper scan. The badge does the bragging, diligence becomes a social object, and nobody has to say the word "slop" out loud.
gift a scan →"Mmm — that's a good scrabcake."
— a satisfied gifter, tasting someone else's repo
freshly scanned.