Discover gists

LSP proxy workaround for Claude Code on Windows

Makes TypeScript LSP servers work with Claude Code on Windows. Supports both the official typescript-lsp plugin (uses typescript-language-server) and the third-party vtsls plugin from claude-code-lsps (uses vtsls).

Related issues:

#16751 — Node.js spawn() can't execute .cmd wrappers on Windows
#17136 — LSP on Windows

The problems

Bypass Windows 11 Upgrade Assistant / Setup Hardware Checks (TPM, CPU, RAM)

This PowerShell script allows you to bypass TPM 2.0, unsupported CPU, and memory checks enforced by the Windows 11 Upgrade Assistant and setup.exe from Windows installation media. It eliminates common upgrade blocks such as:

This PC doesn't currently meet Windows 11 system requirements.
TPM 2.0 must be supported and enabled on this PC.
The processor isn't currently supported for Windows 11.

What It Does

NB: these instructions are for Ubuntu; they work under WSL on Windows 10 as well

Get the decrypt.sh file from here: https://github.com/zhangyoufu/unifi-backup-decrypt

Make sure prerequisites are installed:

sudo apt update && sudo apt install zip unzip openssl mongo-tools

XS Leaks using disk cache grooming

The admin bot creates a new browsing context with createBrowsingContext() and uses that to create a page. Each browsing context should have a dedicated disk cache but how does chrome handle this? I deduced that it uses in-memory disk cache and it's much smaller than the default on-disk disk cache. The incognito tab of my browser has the same behavior.

The following page alerts "not cached" due to cache miss in incognito mode but no error happens in a regular tab.

$ head /dev/urandom -c 5242880 > chunk
$ cat <<EOF > index.html

	import itertools
	import random
	import time

	A = 0x5deece66d
	B = 11
	LCG_MOD_POW = 48
	M = 2**LCG_MOD_POW
	BIG_MASK = M - 1

	from concurrent.futures import ThreadPoolExecutor
	import json
	from pathlib import Path
	import sys
	from typing import List

	import requests

	### How to extract the AUTHTOKEN:
	### 1. Open DevTools (f12 on chrome)

	<?php
	/*
	+----------------------------------------------+
	\| \|
	\| PHP apache log parser class \|
	\| \|
	+----------------------------------------------+
	\| Filename : apache-log-parser.php \|
	\| Created : 21-Sep-05 23:28 GMT \|
	\| Created By : Sam Clarke \|

	# weapon_cfg_singlefile.py
	import struct, mmap
	from collections import namedtuple
	from array import array

	Weapon = namedtuple("Weapon", "id name ammo fire_rate weapon_type sound damage")

	# Header: magic(4) 'WCFG', version(1), 3 reserved bytes, count(uint32), names_size(uint32), sounds_size(uint32)
	_HEADER_STRUCT = struct.Struct("<4sB3xIII")
	_HEADER_SIZE = _HEADER_STRUCT.size

	#!/usr/bin/dotnet fsi

	open System
	open System.Collections.Generic
	open System.Globalization
	open System.IO
	open System.Net.Http
	open System.Numerics
	open System.Text

	"""
	The most atomic way to train and run inference for a GPT in pure, dependency-free Python.
	This file is the complete algorithm.
	Everything else is just efficiency.

	@karpathy
	"""

	import os # os.path.exists
	import math # math.log, math.exp