Juan Simón juanchoperezj

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.

#!/bin/bash

Claude Code statusline with two-line design and semantic color theory

Line 1: Repo/code context (cool blues/cyans)

Line 2: Session/context info (complementary warm tones)

input=$(cat)

Extract data from JSON

Code Review Checklist

Overview

Comprehensive checklist for conducting thorough code reviews to ensure quality, security, and maintainability.

Don't run tests or linting, this is done automatically. Just inspect the code.

Review Categories

1. Use

1.1. in callback

const handlePress = useCallback(() => {
	// handle press
}, [])

Good 👍

<Button onPress={handlePress} />

	import React from "react";
	import {
	View,
	SafeAreaView,
	useWindowDimensions,
	TextInput,
	} from "react-native";
	import Animated, {
	Extrapolate,
	interpolate,

	import { BottomSheetModalProvider } from "@gorhom/bottom-sheet";

	/**
	* This gist describes how to build a global alert system that follows a similar
	* API as RN's Alert.alert().
	*
	* This can be cleaned up and improved, especially in the GlobalAlertManager
	* side, but it's a good starting point.
	*/

	export const toCamelCase = (e) => {
	return e.replace(/_([a-z])/g, (g) => g[1].toUpperCase());
	};

	export const toSnakeCase = (e) => {
	return e.match(/([A-Z])/g).reduce(
	(str, c) => str.replace(new RegExp(c), '_' + c.toLowerCase()),
	e
	)
	.substring((e.slice(0, 1).match(/([A-Z])/g)) ? 1 : 0);

	import { createAction } from 'redux-actions';
	import {
	NAVIGATION_PUSH,
	NAVIGATION_POP,
	NAVIGATION_RESET_TO,
	NAVIGATION_POP_TO_ROOT
	} from './actions';

	export const push = createAction(NAVIGATION_PUSH);
	export const pop = createAction(NAVIGATION_POP);

	// set it up
	firebase.storage().ref().constructor.prototype.putFiles = function(files) {
	var ref = this;
	return Promise.all(files.map(function(file) {
	return ref.child(file.name).put(file);
	}));
	}

	// use it!
	firebase.storage().ref().putFiles(files).then(function(metadatas) {