ishida ishida83

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.

claw machine

I'm terrible at controlling actual claw machines, so I coded my own. You can move the claw by holding down the button, and stop by releasing. The main challenge was to chain the animation required to move the claw and the arm. To simulate gravity, the toy tilts based on which part of the toy the claw grabs. The facial expression on the toys and the claw is also animated for fun.

A Pen by Masahito Leo Takeuchi on CodePen.

License.

Based on the sneakyf1shy apple intelligence tutorial v1.7

This will give you the full Apple intelligence model!

Anything besides the renewed siri does NOT work including: Writing Tools, Memories, Reduce Interruptions, Image Eraser, etc.

we (me, cowabunga) will NOT provide official support for this

do NOT ask me/any staff members for help

the sneakyf1shy apple intelligence tutorial v2.0

Warning

This is patched as of iOS/iPadOS 18.1 DevBeta 5. If you want to follow this, stay on Beta 4.

This actually downloads the models, and is NOT just new SiriUI. Hence, this process is complex and probably not worth it.

⚠️ Prepare to be disappointed and annoyed, and have your time wasted! ⚠️

What does not work: Writing Tools, Memories, Reduce Interruptions, Image Eraser and other tools that are within official Apple Intelligence on supported devices.

Introduction

Apple restricted the access to Xcode LLM (Predictive code completion) feature on China models of Mac. This guide provides a way to bypass that restriction. It's verified on macOS 15.0 Beta (24A5264n), but there is no guarentee that it will always work on later macOS versions.

Prerequisites

Xcode is installed and run at least once.
SIP debugging restrictions are disabled (via csrutil enable --without debug command in recovery mode).

Disclaimer

安装 docker-ce

# https://developer.aliyun.com/article/110806
sudo apt-get update
sudo apt-get install ca-certificates curl
sudo install -m 0755 -d /etc/apt/keyrings
sudo curl -fsSL https://mirrors.aliyun.com/docker-ce/linux/debian/gpg -o /etc/apt/keyrings/docker.asc
sudo chmod a+r /etc/apt/keyrings/docker.asc

Compiling a standalone executable using modern JavaScript/TypeScript runtimes

We have the same code working using node, deno, and bun.

E.g.,

bun run index.js

	## Workflow Orchestration

	### 1. Plan Mode Default
	- Enter plan mode for ANY non-trivial task (3+ steps or architectural decisions)
	- If something goes sideways, STOP and re-plan immediately
	- Don't keep pushing.
	- Use plan mode for verification steps, not just building
	- Write detailed specs upfront to reduce ambiguity

	### 2. Subagent Strategy

	<div style="position:absolute;top:-999px;left:-999px">
	<svg
	id="effectSvg"
	width="200"
	height="200"
	viewBox="0 0 200 200"
	xmlns="http://www.w3.org/2000/svg">


	<filter id="displacementFilter4">

	let totalSize = 0;

	for (let key in localStorage) {
	if (localStorage.hasOwnProperty(key)) {
	let keySize = new Blob([key]).size; // Size of the key
	let valueSize = new Blob([localStorage[key]]).size; // Size of the value
	totalSize += keySize + valueSize;
	}
	}