Vibe Check
is-vibecoded.vercel.app- Category
- Developer Tools
- Rank
- No. 4244Tools index
Previous survey · No. 4023 ·
- Pricing
- Free
- Type
- APP
- Use case
- Data Processing
- Interfaces
- Web
- Date
About
A browser interface that scores coding-pattern heuristics for pasted code or a website URL. It attempts external AI analysis; the tested fallback result did not establish whether a human or AI wrote the code.
At a glance
Vibe Check tries to label pasted code or a website as AI-generated or human-written. Skip it as evidence of authorship: in the tested degraded path, two synthetic samples received the same 30% HUMAN CODED result despite zero detected signals. Its useful role is limited to curiosity about coding-pattern heuristics, and private code should stay out of it.
Good fit
Someone exploring what a code-pattern detector chooses to count, with a disposable example and no consequential decision attached. The interface reports matched AI and human signals and a local score. That can illustrate a heuristic, but a plausible-looking percentage should not be read as a calibrated probability that a person wrote the code.
Consider something else when
You are deciding whether to trust a contributor, accuse someone of using AI or disclose proprietary source. The captured client code attempts to send a code excerpt to an external model provider, contrary to the page’s blanket claim that analysis runs entirely in the browser. The URL mode also contains external fetching proxies; it was not exercised.
Start here
If you still want to explore it, use a small synthetic snippet in Paste Code and read the signal counts and analysis status before the gauge. A three-line HTML example produced zero AI and zero human signals with an unavailable-model error. Adding only a Generated by ChatGPT comment left the same result. Neither output established the code’s origin.
Cost and requirements
The public pasted-code trial required no sign-in or payment. The code path attempts external AI analysis, so browser access alone does not imply local processing. The tested model call reported that the configured model did not exist or was inaccessible; that error left a local analysis result on screen.
Limits to know
The client sets the normalized score to 30 when no patterns match, and the gauge displays the local heuristic. That explains these observed results without establishing broader accuracy. Two samples are not an evaluation dataset, and this guide does not estimate a false-positive rate. There is no basis here to use the tool for an authorship judgment.
Sources checked 2026-10-30. Source-based assessment.
What it can do
Inspect local AI and human coding-pattern matches
Pasted code → Matched signals and a heuristic score
Attempt external AI analysis
Code excerpt and local statistics → AI explanation or a provider-unavailable warning
Why it made the leaderboard
A code-pattern classification experiment whose limitations matter more than its gauge: the tested zero-signal result defaulted to 30% HUMAN CODED, which does not establish authorship.
Intel on Vibe Check
Tags
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.