MindFry

A Subjective Biological Memory Substrate

Memory = f(access, decay, mood, bonds)

---

⚠️ EXPERIMENTAL: MindFry is in active R&D. The API is volatile. It simulates biological memory processes which may result in data inhibition based on the system's "mood". Do not use for banking.

---

What Makes This Different?

Traditional databases are objective — they store exactly what you give them.

MindFry is subjective — it processes data through a cognitive layer that can:

• Forget data that isn't accessed (organic decay)
• Suppress data it finds antagonistic (mood-based inhibition)
• Strengthen frequently accessed data (Hebbian learning)
• Propagate stimulation through neural bonds (synaptic chains)
• Remember how it died and adapt accordingly (crash recovery)

Key Features

Tri-Cortex Architecture

Decisions use Balanced Ternary Logic (Setun):

Value	Meaning
+1	True / Excitation
0	Unknown / Neutral
-1	False / Inhibition

Mood & Personality

The database has a Personality Octet (8 dimensions) and a Mood that affects data prioritization.

• High Mood → More memories feel accessible
• Low Mood → Only important memories surface
• Override: Use BYPASS_FILTERS flag for guaranteed access

Synaptic Propagation

A (+1.0) → B (+0.5) → C (+0.25) → ... (damped)

Touch one memory, its neighbors tremble.

Stability Layer (v1.7+)

Production-grade resilience:

• Crash Recovery — Detects shock (unclean shutdown) and coma (prolonged downtime)
• Warmup Enforcement — Rejects operations during resurrection (Ping/Stats exempt)
• Exhaustion Backpressure — Circuit breaker under high load
• Graceful Shutdown — Pre-shutdown snapshot with marker

Sparse Snapshots (v1.9+)

Portable memory with minimal overhead:

• Sparse Serialization — Only non-empty engrams saved
• zstd Compression — 10-20x size reduction
• 1.5GB → ~100 bytes for empty arenas

Quick Start

Docker (Recommended)

docker run -d -p 9527:9527 ghcr.io/cluster-127/mindfry:latest

From Source

git clone https://github.com/cluster-127/mindfry.git
cd mindfry
cargo run --release --bin mindfry-server

# In another terminal
cargo run --bin mfcli -- ping
cargo run --bin mfcli -- create fire 0.9
cargo run --bin mfcli -- stimulate fire 1.0
cargo run --bin mfcli -- stats

Cargo

cargo add mindfry

SDK

npm install mindfry

import { MindFry } from 'mindfry'

const brain = new MindFry({ host: 'localhost', port: 9527 })
await brain.connect()

// Touch one memory...
await brain.lineage.stimulate({ key: 'trauma', delta: 1.0 })

// ...and its neighbors tremble
const associated = await brain.lineage.get('fear')
console.log(associated.energy) // Increased by propagation

Status

Component	Status
Core Engine	✅ Stable
SDK (TypeScript)	✅ v0.4.0
Persistence (sled)	✅ Stable
Auto-Propagation	✅ Stable
Stability Layer	✅ v1.7+
Sparse Snapshots	✅ v1.9+
Documentation	✅ Live
OQL (Query Language)	🚧 v2.0
CEREBRO (GUI)	💭 Planned

License

Apache-2.0 © Erdem Arslan

---

Core Formula: E(t) = E₀ × e^(-λt) × (1 + Σ stimulation)