ml-systems-lab runs commands on machines you point it at: it copies an agent over SSH, executes it with kubectl exec, and launches local inference binaries. A config file is therefore code, and should be treated as such: run configs you wrote or have read, the same way you would a shell script.

Please report security problems privately, through GitHub's private vulnerability reporting for this repository, rather than in a public issue. Include the version (mlsys --version), the command or config, and what an attacker controls.

This is a single-maintainer project, so expect an acknowledgement within days rather than hours. Fixes are released as a new PyPI version and noted in CHANGELOG.md.

Only the latest release on PyPI receives fixes.