See what's exposed in your Supabase policies, live site, .env file, git history and Firebase rules.
Tuvio reads the repo your assistant wrote, flagging open Supabase RLS, leaked keys, exploited packages and mobile config mistakes.
Your repository or your live site, free to scan. No account, no card.
Security findings
4 things here can be used against you as the code stands today.
- CriticalAnyone signed in can read every user's datasupabase/migrations/0002_profiles.sql
- CriticalAnyone can fake a successful paymentapp/api/webhooks/stripe/route.ts
- HighThe admin check only runs in the browserapp/api/admin/users/route.ts
- HighA secret key may be hardcoded in the codelib/cache.ts
Easy to ship. Hard to spot.
One customer logs in and reads every other customer's rows.
Supabase row-level security
Your service-role key ships to every browser.
NEXT_PUBLIC_, VITE_, EXPO_PUBLIC_
The admin check only exists in the page.
"use client" components
Change the ID in the URL, read someone else's record.
API routes
A server action writes to your database for anyone.
"use server" functions
Anyone can fake a "payment succeeded" and get your paid plan free.
Stripe webhooks
The key you deleted is still in git history.
Every commit, not just the latest*
A package in your app has a flaw attackers are already exploiting.
Your lockfile
Every scan also covers other known vulnerabilities in your dependencies, Docker and infrastructure config, and injection and crypto mistakes across the code.
See what it catches in your stack
All coverageWhy not just ask your assistant?
- You deleted the key. Git kept it.
- Your assistant reads the files you have now. Anyone who clones your repo can read the history. We scan it.*
- You have to know what to ask.
- "Check my security" gets you a generic list. We check what you've never heard of: your Firebase rules, your Android signing keys, the secrets in your Expo config.
- Ask twice, get two lists.
- Same commit, same findings, every time. When one disappears, you fixed it.
- It doesn't know last month's news.
- An assistant knows about broken packages up to the day it was trained. We check yours against a list we update every week.
- It won't check again next week.
- An assistant answers when you ask. Solo scans your repo again every week and emails you when a new High or Critical issue turns up.
Don't take our word for it. Ask your assistant first, then run a free scan on the same repo and compare. What an assistant is good at is judging what we found — that's AI Review.
The full answerThen AI Review tells you which ones are real.
It comes with Solo, $19.99 a month. It drops the ones that aren't real, puts the rest in order, and writes the fix as a prompt for Cursor or Claude Code.
A secret key may be hardcoded in the codelib/cache.ts
Not a blocker. The value is a cache label, not a password.
In supabase/migrations/0002_profiles.sql the row-level security policy on profiles uses USING (true), which lets any authenticated user select every row. Write a new migration that drops that policy and creates one restricted to the owner (auth.uid() = user_id). Leave the insert and update policies alone. Then add a test that signs in as a second user and asserts they cannot select the first user's row. Do not disable RLS to make anything pass.
Ask your assistant. It asks Tuvio.
Connect your AI tool once. Then ask from your project folder in plain sentences: it starts the scan, reads what Tuvio found and works through the fixes with you, in your editor.
- It scans what you pushed.
- Tuvio reads the branch on GitHub, not your local changes.
- It asks before it spends.
- Fixes come from an AI review, and your assistant asks first. Your AI only spends review credits you already have, and only after you switch this on.
- It gets findings, not a copy of your repo.
- Leaked-key findings only once Tuvio confirms the repository is yours.
Works over MCP with Claude Code, Cursor, Codex or others. Included with a free account.
Connect your AI- You
- Scan this repo with Tuvio
- Assistant
- Tuvio scanned what you pushed. 4 issues block launch:
- Critical Anyone signed in can read every user's data
- Critical Anyone can fake a successful payment
- High The admin check only runs in the browser
- High A secret key may be hardcoded in the code
- You
- Fix the security issues Tuvio found
- Assistant
- The fixes come from an AI review, which uses 1 of your reviews. Start it?
- You
- Yes
- Assistant
- The review confirmed 3 and dropped 1 that isn't real. Starting with “Anyone signed in can read every user's data” in
supabase/migrations/0002_profiles.sql.
A clean report is true for one commit.
You'll keep prompting, and packages you already use will keep turning up on the exploited list. Solo scans your repositories again every week and emails you only when something new and serious appears. Nothing new, no email.
- It runs every week without you.
- Up to 5 GitHub repositories, public or private. A private one needs a read-only access token.
- It comes with AI reviews.
- Solo also includes 20 AI reviews a month.
- You only hear about what's new.
- The first scan sets the baseline. After that, an email means a High or Critical issue that wasn't there last week.
- The email leads to the fix.
- It links to the report, where AI Review confirms what's real and writes the fix prompts.
- Week 1 First scan done
- Week 2 No new High or Critical in what it checked
- Week 3 No new High or Critical in what it checked
- Week 4 1 new High or Critical, emailed
1 new High issue in example/demo-shop
The latest scan of example/demo-shop found issues we have not seen in it before.
- Vulnerable dependency
- 1
What happens to your code.
- It is read, never run.
- Nothing is built or installed, so no
postinstallscript from your repo, or from anything it depends on, runs on our machines. - AI Review sees findings, not your whole repository.
- Each finding and the few lines of code it points to. Only when a review runs, sent to Anthropic, with any credential masked before it leaves.
- Delete means delete.
- Closing your account removes projects, scan history and repository tokens.
- The copy is deleted when the scan ends.
- We clone the repository, scan it and delete the clone. The few lines each finding points to are kept for 30 days so AI Review can read them, then deleted.
A clean report means these tools found nothing blocking on that commit. That is a good sign and we will never call it a guarantee. Tuvio never edits your code or opens pull requests. A connected AI tool makes the changes you ask for, in your editor. Accessibility is out of scope because testing it means running your app, so it has its own tool at codexmotive.com/audit.
Find out tonight, before your users do.
Free to scan. No card. Compare plans