Joshua Levy jlevy

memory_guard.py

A last-resort supervisor that watches a process tree on macOS and sheds it before the host becomes unusable. Standard library only, no dependencies, one file.

For workloads that fan out into many short-lived, memory-hungry child processes — agent CLIs, parallel builds, batch jobs — where nothing upstream bounds how many run at once. Where a pool offers real admission control, that pool should do this job and this tool should never fire.

Logical Word Count

Logical word count is a cheap, deterministic, tokenizer-free measure of text length that stays stable across natural languages, code, math, and mixed content, and stays roughly proportional to LLM token counts — while remaining equal to ordinary word count for plain English prose.

It is defined in two rules:

Wide (East Asian) characters count as ½ logical word each. This covers Chinese

Standard Documentation Guidelines

Version: v0.3 (last update 2026-05-05)
Joshua Levy (github.com/jlevy)

Purpose

Both agents and humans benefit from accurate, maintained documentation. These are brief and general guidelines for humans and agents when writing and organizing code, text files, and documentation.

Pure ESM package

The package that linked you here is now pure ESM. It cannot be require()'d from CommonJS.

This means you have the following choices:

Use ESM yourself. (preferred)
Use import foo from 'foo' instead of const foo = require('foo') to import the package. You also need to put "type": "module" in your package.json and more. Follow the below guide.
If the package is used in an async context, you could use await import(…) from CommonJS instead of require(…).
Stay on the existing version of the package until you can move to ESM.

Overview

Problem

If you use Segment's Google Analytics server-side integrations, even if you follow all of their documentation / recommendation, it is impossible to properly track your visitors/sessions in GA from anonymous through identified.

This means that if you use Segment like this, you cannot do very basic things in GA like understanding attribution of your product's sign ups. In GA, a brand new session is created for the identified users, which is not correct.

This problem applies to using Segment on the web with analytics.js when the Google Analytics Destination is set to Cloud Mode, or when using a true server-side Source such as Python/Ruby/Node.

ANSI Escape Sequences

Standard escape codes are prefixed with Escape:

Ctrl-Key: ^[
Octal: \033
Unicode: \u001b
Hexadecimal: \x1B
Decimal: 27

	#!/usr/bin/env -S uv run --script
	# /// script
	# requires-python = "==3.13"
	# dependencies = [
	# "requests>=2.32.3",
	# "rich>=14.0.0",
	# "markdownify>=1.1.0",
	# "readabilipy>=0.3.0",
	# ]
	# ///

	// These hashes are for algorithmic use cases, such as bucketing in hashtables, where security isn't
	// needed and 32 or 64 bits is enough (that is, rare collisions are acceptable). These are way simpler
	// than sha1 (and all its deps) or similar, and with a short, clean (base 36 alphanumeric) result.

	// A simple, insecure 32-bit hash that's short, fast, and has no dependencies.
	// Output is always 7 characters.
	// Loosely based on the Java version; see
	// https://stackoverflow.com/questions/6122571/simple-non-secure-hash-function-for-javascript
	const simpleHash = str => {
	let hash = 0;

	#!/bin/bash
	# Convenience script to pretty-print and browse colored JSON without remembering flags.
	# Read from files or stdin.
	jq -C . "$@" \| less -R -m

	#!/bin/bash

	set -euo pipefail

	file1="${1:?

	Usage: $(basename $0) file1.json file2.json

	Show diff of JSON, normalizing key ordering.
	}"