karpathy

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.

tanaikech

Understanding Flow of Request to Web Apps Created by Google Apps Script

Here, I would like to introduce a report for understanding the flow of the request to Web Apps created by Google Apps Script. There might be a case that various applications using the Web Apps are created and the Web Apps are used as the webhook. In that case, it is considered that when you have understood the flow of requests to the Web Apps, your goal might be able to be smoothly achieved. In this report, I would like to introduce the information about it.

Sample situation

As a sample situation, the sample script for the Web Apps is as follows. And, please set a Spreadsheet ID to work_. In this sample, the data from the request is put into the Spreadsheet. And, the event object is directly returned from the Web Apps.