“The same security principles applied to CI/CD tools like Jenkins, such as creating strict security boundaries for agents, should also be applied to generative AI agents to mitigate risks.”