yof by sōtem · Early access
Your AI built it. yof checks it, explains it, guards it and watches it.
yof: your code's best friend. Catches leaked keys, open databases and risky changes before you ship. Free for your own apps, private, any AI.
Fix nowAnyone can read your customers' emails
Your customer list is open to the public. Anyone who finds your app's address can download every email in it.
Turned on row level security. Each person can now read only their own row.
Checked again. The table is closed. Nothing else changed.
Fix nowAnyone can read your customers' emails
Your customer list is open to the public. Anyone who finds your app's address can download every email in it.
Turned on row level security. Each person can now read only their own row.
Checked again. The table is closed. Nothing else changed.
Preview: how yof will work
AI writes fast. Mistakes ship faster.
An assistant can build a whole app in an afternoon. It can also leave a door open without telling you. yof follows every change and brings it back to one clear answer: safe to share, or what to fix first.
01 Catch
Find the open doors before someone else does
yof will look at your app the way a stranger would, and list what is exposed: keys, customer data, pages without a lock.
Fix nowAnyone can read your customers' emails
Your customer list is open to the public. Anyone who finds your app's address can download every email in it.
Turned on row level security. Each person can now read only their own row.
Checked again. The table is closed. Nothing else changed.
02 Explain
Plain words, and a fix you can paste
Every finding will say what is wrong, why it matters to you and how urgent it is. Then yof writes the fix as a prompt for the AI you built with.
Findings
Rules
03 Guard
A guard that follows your rules
yof is an agent with a rules file. It can comment on, warn about or block a risky change, and it never pushes, merges or deletes anything on its own.
This change lets anyone delete accounts.
Suggested change. yof never commits it for you.
04 Watch
Still watching after launch
Once your app is live, yof will check it every day and send one short email: all good, or the one thing to fix.
Any AI
Use the AI you like, or none at all
.-"""-.
/| o o- - |\
\| ^ |/
\ '-' /
'---'For developers
The same checks, from your terminal
A command, a commit hook and a pull request check. SARIF for code scanning, a baseline for old projects, and pull requests from forks never see your secrets.
The install line arrives on launch day. Until then there is nothing to install, so there is nothing to copy.
Before every commit
A quick key check runs before code leaves your computer, and stops the commit if a key would leak.
On every pull request
One comment that updates in place, with a block only when your rules say so.
In your own CI
SARIF for code scanning and a baseline for old projects, so only new problems fail the build.
Promises
What yof promises
Runs on your side
On your computer or in your own CI. Your code goes only to the AI you pick.
No telemetry
yof will not report on you or your project to anyone, us included.
Keys stay hidden
Any key yof finds is hidden before anything is sent to an AI.
Open source (Apache 2.0)
The code goes public on launch day, so anyone can read exactly what it does. Free for your own apps. Team dashboard and history in sōtem.
Only reads public pages
On a live site it never attacks, logs in or floods anything.
Honest about limits
It can't prove an app is completely safe. It says what it checked and what it didn't.
Hear the day yof is ready
yof is still being built, so there is nothing to install yet. Leave your email and we'll write once, when you can try it.