Skip to content

Security: CodeLaser/maddi

Security

SECURITY.md

Security policy

Reporting

Mail bart.naudts@codelaser.io, or open a private advisory. Please do not open a public issue for something exploitable.

You will get an acknowledgement within a few working days. maddi is developed by one person, so please read the scope below before assuming urgency — most of it does not apply.

What maddi is, and what that means for its attack surface

maddi is a static analyzer: a developer tool that reads source and class files and reports annotations. It opens no ports, serves no requests, and stores no credentials. There is no deployment to compromise.

That said, it does two things worth taking seriously:

  • It parses untrusted input. Running maddi over a hostile codebase, jar or build log should not do anything worse than fail. A crash is a bug; anything that reads or writes outside the configured input and output directories, or executes code from the analyzed project, is a security issue — report it privately.
  • The IDE daemon listens locally. The maddi-ide-daemon process serves the IntelliJ, Eclipse and VS Code front ends. It is intended for local use by the developer running it. A way to reach it from off-machine, or to make it act on input from another user on the same machine, is in scope.

Out of scope: findings that require an attacker who already runs code as you, and reports about the analyzed project's vulnerabilities rather than maddi's own.

Supported versions

maddi is not yet production ready and has no release branches. Fixes land on main; the annotations artifact (io.codelaser:maddi-support) is versioned on Maven Central. Please report against main or the latest published version.

There aren't any published security advisories