Don't open a public issue for a security problem. Use GitHub's private reporting instead: go to the repository's Security tab and choose Report a vulnerability (Private Vulnerability Reporting). That opens a private thread with the maintainers.

Tell us what you found, how to reproduce it, and what an attacker could do with it. A proof of concept helps. We'll confirm we received it and keep you posted while we work on a fix.

OmegaOS runs AI agents with real shell access on a machine you control, and it stores credentials under ~/.omega/. The things we care most about:

• Credentials (~/.omega/, provisioning groups) leaking into logs, the repo, or another project's scope.
• An agent or worker escaping its declared file scope, or one session reading another's secrets.
• The Telegram bridge accepting commands from someone outside the configured chat or sender allow-list.
• Command injection through a dispatched mission, a session name, or a config value.

The agent runtime executes code on purpose. Running a worker that modifies files in its working directory, or an oracle dispatching shell commands, is the tool doing its job, not a vulnerability. Authorized security testing (the /hack pipeline) is a feature.

OmegaOS is pre-1.0. Only the latest commit on main is supported. There are no backports yet.