AI agents that can merge PRs, deploy to Kubernetes, and write to production databases are becoming mainstream, but without safeguards they pose a risk. In response, developer Alex has released Cordum, an open‑source governance layer that forces every action through a policy check before execution.

Security teams have repeatedly blocked AI deployments because they lack audit trails and approval workflows. Cordum’s Safety Kernel evaluates policies—allow, deny, require approval, or throttle—before any job runs, ensuring that destructive operations are caught and that human oversight remains in the loop for every deployment.

Built in Go, NATS, and Redis, Cordum delivers sub‑5 ms policy evaluation and can process over 10,000 events per second. Its CAP protocol lets workers written in any language—Go, Python, Node, or C++—communicate with the control plane, making integration painless for enterprise deployments across cloud infrastructure.

With 16 pre‑built packs for Slack, GitHub, Kubernetes, and more, Cordum lowers the barrier to secure AI automation. The roadmap includes a Helm chart, a managed Cordum Cloud offering, and a visual workflow editor, positioning it as a future‑proof solution for regulated industries in 2025.