yoniLavi

Persistent Experience for Language Model Agents

A research proposal on hybrid memory architectures

Budget envelope: $100,000 in cloud compute

Timeframe: ~6 months of focused work

Date: May 2026

---

yoniLavi

#!/usr/bin/env bash
# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# claude-shield — AI-powered permission hook for Claude Code
# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
#
# Flow:  AllowList (settings.json, instant) → DENY regex (~40ms) → LLM review (~1-2s)
#
# Safe commands should live in permissions.allow in settings.json so
# they never reach this hook. Only unknown/risky actions arrive here.
#

yoniLavi

"""
Discrete probability distribution simulator with proper dependency tracking.

This module implements discrete probability distributions from first principles,
correctly handling complex dependencies between random variables.

Examples:
    >>> die = RandomVariable.die()
    >>> coin = RandomVariable.coin()

yoniLavi

Thought about cryptic crossword clues for 7m 4s:

Laying out the options I’m working through a cryptic crossword clue, analyzing synonyms, and considering anagrams and wordplay to uncover the answer.

Identifying an anagram OK, let me see. I’m thinking the cryptic clue suggests “OXTEMPTSSHARK” is an anagram. Trying different letter arrangements to find a common (1,5,3,4) phrase.

Identifying potential words I’m gathering info on the frequency of each letter. Using “A” as the starting letter, I’m considering a (1,5,3,4) phrase with available letters.

yoniLavi

yoniLavi

#include <Wire.h>
#include <LiquidCrystal_I2C.h>

LiquidCrystal_I2C lcd(0x27, 16, 2);

const int trigPin = 9;
const int echoPin = 10;
const int ledPin = 3;  // PWM pin for LED

// Adjust these values to match your desired range

yoniLavi

// Potentiometer
int potPin = A0;

// 7-Segment Display
const int sA = 1;
const int sB = 7;
const int sC = 5;
const int sD = 4;
const int sE = 3;
const int sF = 2;

yoniLavi

import pulp
import pandas as pd


warehouses = pd.Series({"Dublin": 300, "Cork": 200, "Galway": 200})
targets = pd.Series({"Belfast": 150, "Limerick": 250, "Sligo": 200})
costs = pd.DataFrame({
    "Belfast": [16, 14, 13],
    "Limerick": [18, 12, 15],
    "Sligo": [11, 13, 17]

yoniLavi

from controller import Robot, DistanceSensor, Motor

TIME_STEP = 64
SPEED = 3.14
PROXIMITY_THRESHOLD = 80

robot = Robot()

def setup_sensor(device_id):
    sensor = robot.getDevice(f"ps{device_id}")

yoniLavi

from sympy import symbols, Eq, laplace_transform
from sympy.physics.mechanics import dynamicsymbols

t, s, G, F = symbols('t s G F')
g, f = dynamicsymbols('g f') # type: ignore

# Equation: d^2g/dt^2 + 4dg/dt + 3g = 0.2 df/dt + 1.5f
eq_lhs = g.diff(t, t) + 4 * g.diff(t) + 3 * g # type: ignore
eq_rhs = 0.2 * f.diff(t) + 1.5 * f # type: ignore