RatOrNot
All guides

Understanding results

False positives in jar scanners

Last updated November 16, 2026

Mods do powerful things, and some harmless ones look a lot like malware on the surface. A flag is a reason to read the evidence, not proof of malice. Knowing why legitimate mods get flagged helps you read a scan result sensibly.

Why false positives happen

  • Update checkers make web requests and may download a new version. They resemble a loader at a glance.
  • Compatibility layers use reflection to call other mods, which is also how obfuscated malware hides its calls.
  • Shaded libraries such as HTTP clients, Gson or Netty bundle many networking and crypto classes that the mod never uses.
  • Obfuscation (including ProGuard shrinking) makes code harder to read, which can look evasive even when the author just wants a smaller jar.
  • Screenshot, clipboard and account features. A screenshot mod uses the same API as spyware; a mod that displays your username reads the same session object a stealer does.

Signals that matter

A single suspicious-looking call rarely means much. Combinations do: reading an access token and sending it to a webhook, or downloading bytes and loading them as a class, or touching browser profile folders from a game mod. What the data is matters more than which call moves it.

A rough guide to weighing a finding:

  • Weak on its own: a network call to the mod's own update API, reflection on other mods, a shaded library, a shrunk jar.
  • Worth a closer look: a network call to a host that has nothing to do with the mod, deleting files, delayed start, hidden strings.
  • Serious: credentials, session tokens or browser and wallet files reaching a network call; downloaded bytes becoming code; writing to startup locations; modifying other jars.

How RatOrNot handles it

RatOrNot scores behaviour in context. It follows data from where it is read to where it goes, and lowers the weight of signals that are explained by a benign pattern, such as showing a username in a menu or contacting a mod-hosting API for updates. Results list the evidence so you can verify it. Each rule is described on what RatOrNot detects.

How to check a flag yourself

  1. Open the finding and note the class and method it names.
  2. Ask whether the behaviour fits the mod. A map mod calling a screenshot API is plausible; a performance mod reading Discord storage is not.
  3. Check where the address in the evidence leads. A mod's own API or Modrinth is normal; a random webhook is not.
  4. Compare the file's hash and size with the official download.
  5. Look at the author, source repository and release history.

If you think a result is wrong

Check the evidence, compare against the author's official download and tell us if a legitimate mod was flagged. Include the file hash and what you verified. Reports from authors and players help improve the rules. If you are a mod author whose project was flagged, send the link to your public source and we will review it.

The opposite error also exists: a clean result does not prove a file is safe. See how RatOrNot works for the limits of static analysis.