cgroup-v2

Sandbox service built on Linux container technologies with simple REST and gRPC API

CLI Noise-aware CPU pinning

Ubuntu 24.04 XFCE desktop container — systemd PID 1, VNC + xrdp + SSH, optional NoMachine NX and NVIDIA GPU. Multi-Python (3.10/3.11/3.12/3.13), Firefox, VS Code, OBS, VLC. Non-privileged on cgroup v2.

Per-app network rate limiter and traffic monitor for Linux. Pure eBPF. Boring and silent but killer.

A Linux container runtime built from scratch in C++20 — namespaces, OverlayFS, cgroup v2, and dual-stack networking without Docker or any container framework.

Lightweight Prometheus exporter for Docker container metrics — built for ARM64 and cgroup v2

A lightweight cross-platform process supervisor for embedded devices and edge nodes, with dependency orchestration, automatic recovery, and whole process-tree management.

Memory, CPU and I/O limits for your own Linux apps, without root, plus a guard that pauses a runaway app before the desktop freezes.

Terminal-first Kubernetes memory inspector for evidence-backed cgroup v2 diagnosis across Pods, containers, workloads, and nodes.

Tool-call-granularity CPU/memory characterization and control for AI coding agents, built on cgroup v2 and eBPF.

VRAM protection broker for OpenRC using DMEM low

pmdaemon service

Kotlin async child-process management for tokio: whole-tree kill-on-drop (no orphans), plus streaming, pipelines, timeouts, and supervision library.

A lightweight local C++ judge for Linux with process isolation, time/memory limits, and cgroup v2 resource tracking.

A small eBPF-based sandboxing tool designed to block any attempt by a process or its descendants to escalate to root with setuid() / setreuid() / setresuid().

Lightweight Prometheus exporter for Docker container metrics, memory, CPU, network, disk I/O, and container state, without the overhead of cAdvisor.

Sandbox service built on Linux container technologies with simple REST and gRPC API.

Per-process Linux network gate — Little Snitch for Linux, built on Rust + Aya eBPF. Watch outbound connections, allow/deny per process, throttle per-PID bandwidth via tc-bpf.

Pure Ruby interface to Linux Pressure Stall Information (PSI)

A production-oriented BEAM/OTP autonomy kernel for running an untrusted coding worker inside a disposable Linux execution domain while keeping durable authority, policy, effects, verification, recovery, and audit state in a trusted control plane.