A low-level systems governor built natively in Go, bypassing user-space abstractions to directly parse /proc and /sys filesystems (directly interfacing with the Linux Virtual File System).
The architecture tracks cgroups v2 resource delegation (allowing interfacing with the CFS), monitors PSI to detect thread lockups, and isolates physical resident set size (RSS) memory boundaries from virtual memory fragmentation (while actively trapping OOM events).
_WALKTHROUGH
01 / CGROUP_V2_GOVERNANCE
Directly interfacing with the Linux unified cgroup hierarchy - this section dynamically calculates CPU utilization with raw microsecond deltas from cpu.stat, instantly throwing bandwidth limits straight into the cpu.max Virtual File System boundary. Providing control over system resources with zero overhead.
02 / KERNEL_PRESSURE_STALL_PARSING
This function directly parses the Linux kernel's PSI by ripping the raw metrics from the kernel's memory, CPU & I/O subsystems.
03 / CGROUP_INTRUSION_DETSIDE
System metrics are useless if the environment is compromised. This block scans the active cgroup portion for process co-residency anomalies. By continuously scanning cgroup.procs, the daemon can flag any untracked or foreign PIDs that trigger the sandbox boundary.