What shipped, and what shipped the need for it.

The cloudtaser operator now supports automatic TLS certificate rotation for the broker component. Combined with audit logging in the wrapper and Prometheus metrics in the S3 proxy, the full stack now provides production-grade observability and security lifecycle management.

The canonical public demo runs on three real systems — a US GKE confidential-compute target cluster, a beacon relay VM in Frankfurt, and an OpenBao secret store VM in the Netherlands. Self-hosted, no third-party platform: a 4-minute MP4 walkthrough plus a one-driver-many-watchers interactive mode. Every cluster mutation the demo performs is published as YAML at cloudtaser.io/demo-lab/manifests/.

The cloudtaser S3 Proxy now performs full envelope encryption using OpenBao's transient secret engine: unique AES-256-GCM data key per object, wrapped via Transit, plaintext DEK never persists. Compatible with any S3-compatible storage (AWS S3, GCS, MinIO).

The Kubernetes mutating admission webhook now reliably injects the cloudtaser wrapper into pods. Auto-detects container image entrypoints from OCI registries (supports private registries), rewrites commands transparently. No sidecar containers - zero resource overhead compared to sidecar-based injectors. Full CI/CD with integration tests on every release.

cloudtaser's eBPF agent gains BPF LSM hooks as the synchronous deny path for the five highest-risk kernel entry points: ptrace_access_check, file_open (covering /proc/pid/environ and /proc/pid/mem), bpf (unauthorised program loads), socket_sendmsg (secret-buffer exfiltration), and kernel_load_data. By LSM design, the hook fires synchronously before the kernel action completes. On kernels without CONFIG_BPF_LSM=y or where bpf is absent from /sys/kernel/security/lsm (some hardened distros, GKE pre-1.27, AKS Mariner), the agent detects this at startup and falls back to the well-tested kprobe enforcement path. No operator configuration change required. The LSM code merged across cloudtaser-ebpf #183, #185, #190, #193 and is rolling out in the next tagged release; per-kernel runtime confirmation lands once cloudtaser-pipeline#227 unblocks full-matrix BPF verifier validation.

The eBPF agent now attaches to sys_enter_openat tracepoints and kprobes for comprehensive kernel-level monitoring. Detects /proc/environ reads, secret material in network buffers, and translates container PIDs to host PIDs for cross-namespace visibility. Enforcement mode can block unauthorised access in real time.