>_ INSTANCE-05: CHAOS ENGINEERING

A system that has only ever run in comfort has an unknown failure envelope. Chaos Engineering stresses a healthy machine on purpose — cold, breath, voluntary intensity — to prove the hardware can process a severe autonomic spike without crashing. This is hardware benchmarking for the nervous system: the objective is not to cool the machine down, but to expand what it can survive.

> EXECUTION PROTOCOL: Clear the failsafe, then inject thermal load. A 60-second benchmark begins and the stabilization reticle starts drifting on thermal updrafts. Keep it centered inside the Safe Zone — steer with the cursor, a touch drag, or the arrow keys / WASD. Your job is not to stop the burn. Your job is to hold the line while it happens.

> WARNING: Chaos Engineering is for STABLE baselines only. A node already running [MODE 0x02] INCIDENT RESPONSE — redlining, depleted, in recovery — does not need more fault injection; it needs maintenance. The pre-boot failsafe is not a formality. Aborting is a valid execution path.

> TELEMETRY_LEGEND: THERMAL_LOAD is the injected stressor — during a benchmark it pins at 99% and stays there; it is data, not damage. HYPERVISOR_SYNC is your reticle lock on the Safe Zone core. T_MINUS is the benchmark clock. If the reticle stays outside the zone for 3 continuous seconds, the system halts in a Kernel Panic.

> SYS_OPS: Choose your dose: a 30-second Micro-Dose, the 60-second Standard Load, or the 120-second Deep Burn — the turbulence ramps as any window deepens, and the Deep Burn holds peak load for a full minute. Escaping the zone briefly is recoverable; the desync budget refills while you hold the core. [ MANUAL EJECT ] (or press ESC) halts the burn at any time with no penalty — voluntary entry, voluntary exit. Survive the full window and the system logs CAPACITY EXPANDED: the simulated annealing pass recompiles a harder baseline.