teebu

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.

teebu

from random import randint

def generate_fr_vat():
    siren = randint(100000000, 999999999)
    key = (12 + 3 * (siren % 97)) % 97
    return 'FR' + str(key) + str(siren)

print(generate_fr_vat())

teebu

/**
 * Checks ABN for validity using the published ABN checksum algorithm.
 * @author Guy Carpenter
 * @license http://www.clearwater.com.au/code None
 * @param  {String|Number} value abn to validate
 * @return {Boolean} Is ABN Valid
 */
function validateABN (value) {

	var weights = [10, 1, 3, 5, 7, 9, 11, 13, 15, 17, 19],

teebu

##########################################
# Helmfile                               #
# How to manage Kubernetes Helm releases #
# https://youtu.be/qIJt8Iq8Zb0           #
##########################################

# Referenced videos:
# - K3d - How to run Kubernetes cluster locally using Rancher k3s: https://youtu.be/mCesuGk-Fks

#########

teebu

<code_scheme name="Airbnb">
  <option name="RIGHT_MARGIN" value="100" />
  <option name="HTML_ATTRIBUTE_WRAP" value="4" />
  <option name="HTML_ELEMENTS_TO_INSERT_NEW_LINE_BEFORE" value="" />
  <option name="HTML_ENFORCE_QUOTES" value="true" />
  <DBN-PSQL>
    <case-options enabled="false">
      <option name="KEYWORD_CASE" value="lower" />
      <option name="FUNCTION_CASE" value="lower" />
      <option name="PARAMETER_CASE" value="lower" />

teebu

hello

teebu

import numpy as np
import matplotlib.pyplot as plt

from math import exp, log


def getGaussDecayValue(value,origin,scale,decay,offset=0):
    # does this work?
    z = -1*(scale**2)/(2*log(decay))
    a = (-1* (max(0, abs(value - origin) - offset)**2))

teebu

import numpy as np
import matplotlib.pyplot as plt

from math import exp, log


def getGaussDecayValue(value,origin,scale,decay,offset=0):
    # does this work?
    z = -1*(scale**2)/(2*log(decay))
    a = (-1* (max(0, abs(value - origin) - offset)**2))

teebu

PUT mapping_test
{
  "settings": {
    "index": {
      "number_of_shards": "1",
      "refresh_interval": "1s"
    }
  },
  "mappings": {
    "numeric_detection": false, // treat numbers in strings as numbers

teebu

# varnish 4.1
sudo yum -y install autoconf automake jemalloc-devel libedit-devel libtool ncurses-devel pcre-devel pkgconfig python-docutils python-sphinx graphviz
curl -s https://packagecloud.io/install/repositories/varnishcache/varnish41/script.rpm.sh | sudo bash
wget --content-disposition https://packagecloud.io/varnishcache/varnish41/packages/el/6/varnish-4.1.10-1.el6.x86_64.rpm/download.rpm
sudo rpm -Uvh varnish-4.1.10-1.el6.x86_64.rpm
/usr/sbin/varnishd -V


# varnish 4
sudo yum -y install autoconf automake jemalloc-devel libedit-devel libtool ncurses-devel pcre-devel pkgconfig python-docutils python-sphinx graphviz