projects
Projects
Each project is a small working proof of concept you can run yourself, with the code on GitHub and notes on what I learned.
Project 1
Making AI agents safer with an open-source gateway
I put a security checkpoint between an AI agent, the AI model it talks to, and the tools it can use. I tried it on my own computer and checked six areas of security with the NIST framework.
- agentgateway
- MCP
- Kubernetes
- NIST CSF
Project 2
Securing an AI SRE agent on Kubernetes
I let HolmesGPT investigate a Kubernetes incident with read-only access and give the repair to a separate operator agent with its own Keycloak identity. agentgateway, Kubernetes RBAC and network policies keep each agent to its own job.
- HolmesGPT
- Keycloak
- agentgateway
- Kubernetes RBAC