Compare against the kernel
Every reading is checked against what the system itself reports. That habit is how we caught our own memory readout describing occupancy instead of pressure — a number that would have read 98% on any healthy Mac, forever.
We make small, native macOS tools with an unusual obsession: every number they show you is one we can defend. No Electron, no telemetry, no numbers that sound right.
One instrument out now, built to a standard we can keep. More will follow only when they meet it.
Most monitoring software is confident. Ours is checked. Three rules we hold ourselves to, because breaking them is how a tool ends up lying quietly for months.
Every reading is checked against what the system itself reports. That habit is how we caught our own memory readout describing occupancy instead of pressure — a number that would have read 98% on any healthy Mac, forever.
CPU is a CPU-time delta over a fixed window, not a decaying average. Memory is phys_footprint, not resident size. A benchmark whose method is hidden is a marketing number.
A monitor that is expensive to run is a contradiction. Ours states its own footprint, in the app and on this page, and there is a refresh-rate control labelled with its real cost in CPU.
No accounts. No analytics. No crash telemetry. No "anonymous usage statistics". Our tools read your machine and show you the result — the only outbound connections are ones you configure yourself, like a webhook you wrote or a server you own. This is a design constraint, not a setting.