Roja takes action on live infrastructure. That's the whole point — and it's exactly why the security posture is conservative by design.
Roja uses a customer-side credential plane. Device credentials are stored and used inside your network. We don't host them, so we can't lose them.
The agent executes only pre-authorized, verified actions — never arbitrary commands. Risky changes require explicit human approval.
Every dispatch lands in an append-only audit log: detection, diagnosis, action, approval, outcome.
The agent initiates all connections. No inbound ports into your network.
SOC 2 attestation is on our roadmap; we will share specifics once the engagement is funded and scheduled. We'd rather tell you honestly where we are than promise a date we haven't committed to.
Found something? security@getroja.ai. Bug bounty: planned post-launch, not yet live.
