Inspect the whole project
Point VibeTect at the project root, not a single folder. Many issues live in configuration files, middleware and shared helpers rather than the feature you just built.
VTHow it works
No pipeline to configure, no repository to connect and no code to upload. VibeTect works where your project already is — your own computer.
01The workflow
Stage 01
VibeTect is a desktop application, not a website you paste code into. You install it on the computer where your project already lives.
Stage 02
Choose the folder that contains your code — a cloned repository, a project exported from an AI app builder, or a client deliverable.
Stage 03
The analysis runs on your device using your machine’s own resources. Your source code doesn’t need to be uploaded anywhere to be checked.
Stage 04
Findings are grouped by how urgent they are, each with an exact location, a plain-English explanation and fix guidance.
Stage 05
Apply the fix yourself or hand the finding to your AI assistant. Then run the inspection again to confirm the issue is resolved.
02The data boundary
Local-first means the inspection doesn’t depend on sending your source code to someone else’s server. It also means VibeTect only sees what’s in the folder you give it — so be deliberate about what else you check.
03Working with your AI assistant
Vague prompts produce vague fixes. A finding gives your assistant exactly what it needs: the problem, the location and the required outcome.
Fix this security finding. Keep all other behavior the same.
Finding: Any signed-in user can read any invoice (High)
Location: app/api/invoices/[id]/route.ts, line 11
Problem: The invoice is fetched by the ID in the URL without
checking that it belongs to the signed-in user.
Required fix: Scope the query to the signed-in user's ID and return
404 when no matching invoice exists.
After the change, explain what you modified and why.04Getting the most from a check
Point VibeTect at the project root, not a single folder. Many issues live in configuration files, middleware and shared helpers rather than the feature you just built.
Exposed secrets and missing access control are the issues attackers find fastest. Clear those before you polish anything else.
If a key was ever committed or deployed, removing it from the code isn’t enough — it still exists in Git history and any copies. Revoke it and issue a new one.
Database policies created in a web console, storage bucket settings and hosting environment variables don’t live in your folder. Review them in the platform where they’re configured.
A fix can introduce a new problem, especially when an assistant rewrites more than you asked for. A second inspection takes moments and closes the loop.
Run a local security inspection on the project you’re about to deploy — and fix what it finds while it’s still cheap to fix.
Local analysis · No repository access required