Give it your code
Upload a ZIP of your project, or connect GitHub and pick a repository.
Epherem is the free source-code security scanner for apps built with AI tools. Upload a ZIP or connect GitHub; it reads your code without running it and looks for the security mistakes AI-built apps often make, then checks packages with exact versions (from a lockfile) against OSV.dev, a public database of known issues. The full report is free with nothing hidden, and there's no AI in the scan. It reads your code only: it can't see your live site or your database dashboard, and the report names what it couldn't check.
YOUR CODE IS READ, NEVER RUN · NO AI IN THE SCAN · UPLOADS DELETED AFTER EACH SCAN
You give Epherem your code in one of two ways. It checks the upload is safe to open, reads the files that matter, and writes up what it found. Your uploaded code is deleted after the scan; only the report is kept.
Upload a ZIP of your project, or connect GitHub and pick a repository.
ZIP bombs, encrypted archives and files that try to escape their folder are refused.
Installed packages, build output, caches and media are skipped.
55 checks read your code. JavaScript and Python packages with exact versions (from a lockfile) are looked up on OSV.dev, a public database of known issues. Without a lockfile, the report says which ones it couldn’t check.
You get known issues, likely issues, what needs checking, and what the scan couldn't check.
Epherem is built for the web apps AI coding tools produce. It reads some languages better than others, and the report says which parts of your project it analysed. Code it didn't analyse is never reported as clean.
The most checks apply here. Some ways of writing routes can't be read yet; when that happens, the report names the routes it couldn't read.
Checked with less depth. Some checks written for JavaScript don't apply, so fewer problems can be found.
Other languages are named in the report as not analysed. So is anything that isn't in your code.
Each check was written for a mistake that shows up in apps built with AI tools. The report explains every finding in plain words: where it is, what someone could do with it, and how to fix it.
Known issues are packages where a published advisory matches the version you use. Likely issues are where the scanner saw the risky code itself. Findings that need checking are where it looked for a protection and didn’t see it, or where the match depends on context it can’t see. Epherem reads your code but never runs your app, so it can’t confirm a code finding for you.
Epherem is an automated first check, not a penetration test. It works only from your code, so it can't see how your live site behaves or how your services are set up outside the code.
node_modules in. The page leaves it out before uploading, along with build output and other folders Epherem skips.Epherem is free during early access. Sign in with Google or an email code.