I had an old iPhone 7 (A10 Fusion) sitting around and wondered: can I turn this into an authentic web server and hardware control center using zero third-party dependencies?
After jailbreaking via palera1n, I built a pure Python 3 standard library HTTP daemon. The real fun began when I started hooking directly into the physical iOS hardware:
Hardware-Level Exploitation:
- 170Hz Linear Taptic Engine: Calling
AudioToolbox.framework via ctypes so that clicking a button on the web interface physically fires the linear resonant actuator in your palm.
- Dual Stereo Speakers: Synthesizing raw PCM audio waves and piping them through
/usr/bin/afplay to rattle the chassis.
- Apple PMU Telemetry: Querying physical battery current (mA), cell millivolts, and thermistors in real time via
ioreg -rc AppleARMPMUCharger.
- A10 Hurricane Stress Burner: Multi-threaded SHA-256 rounds that push the performance cores, physically warming the 7000-series aluminum shell.
The most critical security rule? Strictly blocking reboot, shutdown, and halt with exit code 13. On a semi-tethered jailbreak, a reboot wipes the kernel exploit from volatile RAM. But sbreload (SpringBoard respring) is fully supported, letting you restart the entire iOS UI without dropping the Python server or losing jailbreak persistence!
On traditional distributions, managing a system over months is an exercise in gradual entropy. You tweak a config in /etc, install some random dependency to test a tool at 2 AM, and six months later you have no idea why an update broke your audio stack or display server.
NixOS completely flips the paradigm. Your entire machine state — kernel modules, systemd daemons, desktop environments, user dotfiles, and packages — is declared in code.
The Generational Safety Net: Every rebuild generates a standalone bootloader entry. If an experimental kernel patch or configuration fails, you don't panic with a Live USB: you reboot, pick the previous generation from GRUB, and you're back running in 10 seconds.
Coupled with Arch for pure cutting-edge desktop testing, Nix is leagues ahead for infrastructure consistency. It appeals directly to the perfectionist part of my brain: everything reproducible, atomic, and version-controlled.
Princeton's FlyWire consortium released the complete connectome of the adult female fruit fly (Drosophila melanogaster) — 49,114 functional neurons and 241,528 biological synapses mapped at nanometer electron microscopy resolution.
Most people analyze connectomes as abstract graph datasets in Python notebooks. I wanted to know: what happens if you embody this connectome inside an agent with physical sensor inputs?
The Neural Plumbing:
- Mapped world illumination to fly photoreceptors (R1–R6, Tm, Mi).
- Tied sweet odorant receptors to sugar cane and flowers; foul receptors to lava, smoke, and fire.
- Fed wall collisions into mechanosensory tactile neurons, and free fall into proprioceptive balance loops.
Running at 20 ticks per second inside a Minecraft player model, the Kenyon cells and dopaminergic plasticity (PPL1 vs PAM) produced real approach and avoidance behaviors purely from biological synaptic weights. It was one of the most rewarding rabbit holes I've ever tackled.
People often imagine cybersecurity as hooded hackers typing green text at 100 WPM. To me, the most fascinating side of security is digital forensics — because it's genuine, methodical detective work.
Attackers can delete event logs, rename malicious binaries, and try to hide their traces, but the physics of computing cannot lie. Memory allocations leave unbacked artifacts, NTFS leaves record signatures in the Master File Table, and packet headers reveal beaconing cadence.
The Rigour of ACPO Standards: In our labs at UCLan, we work under real evidentiary standards. Principle #1 states that no action taken by law enforcement or investigators should alter data held on a computer or storage media. Everything starts with hardware write-blockers, verifiable SHA-256 bitstream copies, and maintaining an unbroken chain of custody.
Taking a raw memory capture from an infected machine and using Volatility 3 to uncover an injected DLL hiding inside an otherwise legitimate svchost.exe process gives you an incredible rush. It’s technical problem-solving at its purest.