Digital Immune System

500 evolved receptors.
3 physical probes.
Zero dependencies.

A Darwinian Kinetic Proofreading organism. Calibrates against its physical substrate, evolves immune receptors through negative selection, detects anomalies at nanosecond resolution. No ML. No training data. One Rust file.

View on GitHubRead the Research
93–95%
Accuracy
Consistent across runs
0–2%
False Positives
Quiet periods
500
Evolved Receptors
Darwinian population
0
Dependencies
Single Rust file
Sensory Organs

Three independent probes

The organism perceives its substrate through three physical channels simultaneously. 50 readings each, median taken. Three numbers define reality at each instant.

Memory
1 MB pointer chase across cache lines. 6 passes per reading. Senses cache hierarchy latency.
Clock
80 consecutive Instant::now() calls. Measures timer access contention under load.
Alloc
8 cycles of 64 KB alloc, touch, drop. Detects allocator pressure and memory subsystem stress.
Architecture

Biology on silicon

Every signal path maps to immunology. Every mechanism has a paper behind it.

Thymic selection
2,000 candidates tested against 60 s of quiet substrate. Reactors to self are killed. 500 survivors form the immune repertoire.
Kinetic proofreading
N consecutive above-threshold signals required to fire. Inspired by McKeithan 1995. Transient noise is filtered by design.
Darwinian evolution
Every 5 cycles: kill low confidence, clone top performers, mutate children. The population adapts in real time.
Regulatory suppression
When the vote declines from a previous block, Treg analog activates. Raises effective quorum. Prevents recovery-phase false positives.
Get Started

Run it yourself

60-second calibration. 100 decision cycles. See for yourself.

terminal
git clone https://github.com/zot-run/zot-cell.git
cd zot-cell
cargo build --release
cargo run --release
python3 analyze.py