Cactus erbanku

import sys
import os

NOP = 0x90
offsets_and_values = {
    0x00030170: 0x00,
    0x000A94D0: NOP, 0x000A94D1: NOP, 0x000A94D2: NOP, 0x000A94D3: NOP, 0x000A94D4: NOP, 0x000A94D5: NOP, 0x000A94D6: NOP, 0x000A94D7: NOP, 0x000A94D8: NOP, 0x000A94D9: NOP, 0x000A94DA: NOP, 0x000A94DB: NOP, 0x000A94DC: NOP, 0x000A94DD: NOP, 0x000A94DE: NOP, 0x000A94DF: NOP, 0x000A94E0: NOP, 0x000A94E1: NOP, 0x000A94E2: NOP, 0x000A94E3: NOP, 0x000A94E4: NOP, 0x000A94E5: NOP, 0x000A94E6: NOP, 0x000A94E7: NOP, 0x000A94E8: NOP, 0x000A94E9: NOP, 0x000A94EA: NOP, 0x000A94EB: NOP, 0x000A94EC: NOP, 0x000A94ED: NOP, 0x000A94EE: NOP, 0x000A94EF: NOP, 0x000A94F0: NOP, 0x000A94F1: NOP, 0x000A94F2: NOP, 0x000A94F3: NOP, 0x000A94F4: NOP, 0x000A94F5: NOP, 0x000A94F6: NOP, 0x000A94F7: NOP, 0x000A94F8: NOP, 0x000A94F9: NOP, 0x000A94FA: NOP, 0x000A94FB: NOP, 0x000A94FC: NOP, 0x000A94FD: NOP, 0x000A94FE: NOP, 0x000A94FF: NOP, 0x000A9500: NOP, 0x000A9501: NOP, 0x000A9502: NOP, 0x000A9503: NOP, 0x000A9504: NOP, 0x000A9505: NOP, 0x000A9506: NO

Codex /goal Implementation

How OpenAI's Codex CLI implements the /goal slash command for persisted long-running task objectives.

The /goal command sets a persisted objective for a long-running task. It's a five-layer system: a SQLite table stores one goal per thread with status (active/paused/budget_limited/complete), optional token budget, and running usage counters. The app-server exposes thread/goal/set/get/clear JSON-RPC methods. The model sees three tools — create_goal, update_goal(complete), and get_goal — but can't pause/resume; those are system-controlled. A runtime event bus hooks into turn lifecycle to track token + wall-clock deltas for accounting, auto-pauses on interrupt, auto-reactivates paused goals on thread resume, and injects budget-limit steering into the model's response stream. The TUI handles the /goal slash command and displays goal state in the status bar.

Author: etraut-openai | 5 PRs, ~15K additions, landed in ~10 days (Apr 16–25, 2026)

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

Conventional Commit Messages

See how a minor change to your commit message style can make a difference.

git commit -m"<type>(<optional scope>): <description>" \
  -m"<optional body>" \
  -m"<optional footer>"

UI Access (UIA)

What's UI Access?

UIA is the way around UIPI ([User Interface Privilege Isolation][1]) which in simple terms is a way of bypassing the security built into Windows to avoid applications interacting with other applications that have a higher integrity level (security).

In other words, when you run a script it cannot communicate with the system or elevated (running as Administrator) processes; this is to avoid insecure and non-sanctioned interactions.

Why would I want UIA?

Problem

Chrome has the feature to automatically add search engine when it detects an input field on websites. After using Chrome months, it often resutls a bunch of search engines stayed in the settings. And the setting page does not provide a convinient way to remove them.

Updates on 2024-11-21

The editting the Web Data as SQLite file did not seem to work anymore. Chrome seems to revert all the changes when restart.

To make it even worse, it seems that the DOM tree is not accessable from the development console on the setting page, so even a bookmarklet could not work. Neighter does Chrome provide an API to access the search engines.

Semantic Commit Messages

See how a minor change to your commit message style can make you a better programmer.

Format: <type>(<scope>): <subject>

<scope> is optional

	<#
	.SYNOPSIS
	Resolves many Windows GUIDs to human friendly values.

	.DESCRIPTION
	Resolve-WindowsGUID.ps1 takes a GUID from a Windows system and attempts
	to return a human friendly value from either a static list or from a
	dynamically generated list of LogProvider GUIDs. There are undoubtedly
	other GUIDs in use throughout Windows that will not fall into either of
	these sets. If you encounter a GUID that you can't resolve via this

	#!/bin/bash
	set -euo pipefail
	trap 'echo "at line $LINENO, exit code $? from $BASH_COMMAND" >&2; exit 1' ERR

	# This is a Claude Code hook to stop it saying "you are right".
	#
	# Installation:
	# 1. Save this script and chmod +x it to make it executable.
	# 2. Within Claude Code, /hooks / UserPromptSubmit > Add a new hook (this file)
	#

	# WSL通过Win访问网络，所以WSL的网关指向的是Windows，DNS服务器指向的也是Windows，设置WSL的proxy为win的代理ip+端口即可
	# WSL中的DNS server在/etc/resolv.conf中查看，该文件是由/etc/wsl.conf自动生成的。
	# 如果关闭了wsl.conf中自动生成resolve.conf并自行修改了resolve.conf，DNS nameserver并不是本机win ip
	# 需要开启wsl.conf的自动生成，再运行以下命令
	# https://zhuanlan.zhihu.com/p/153124468

	# 添加到环境变量设置中，例如~/.zshrc
	export hostip=$(cat /etc/resolv.conf \|grep -oP '(?<=nameserver\ ).*')
	export https_proxy="http://${hostip}:7890"
	export http_proxy="http://${hostip}:7890"