oremj

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.

oremj

https://bouncer-bouncer.stage.mozaws.net/?attribution_code=Y2FtcGFpZ249dGVzdGNhbXBhaWduJmNvbnRlbnQ9dGVzdGNvbnRlbnQmZXhwZXJpbWVudD1leHAxJmluc3RhbGxlcl90eXBlPWZ1bGwmbWVkaXVtPXRlc3RtZWRpdW0mc291cmNlPW1vemlsbGEuY29tJnRpbWVzdGFtcD0xNTkxMjg4NjY3JnZhcmlhdGlvbj12YXIx&attribution_sig=24fe67c6fedbbde3767f9d4ebb73bdcdeb2c7778e042fd429e1b4e5d1b1f3ea2&lang=en-US&os=win&product=firefox-latest-ssl

oremj

date:DATE,time:STRING,location:STRING,bytes:INTEGER,request_ip:STRING,method:STRING,host:STRING,uri:STRING,status:INTEGER,referrer:STRING,user_agent:STRING,query_string:STRING,cookie:STRING,result_type:STRING,request_id:STRING,host_header:STRING,request_protocol:STRING,request_bytes:INTEGER,time_taken:FLOAT,xforwarded_for:STRING,ssl_protocol:STRING,ssl_cipher:STRING,response_result_type:STRING,http_version:STRING,fle_status:STRING,fle_encrypted_fields:STRING,c_port:INTEGER,time_to_first_byte:FLOAT,x_edge_detailed_result_type:STRING,sc_content_type:STRING,sc_content_len:INTEGER,sc_content_start:STRING,sc_range_end:STRING

oremj

#!/usr/bin/env python3

import click
import boto3
import time

logs_client = boto3.client('logs')


@click.command()

oremj

Keybase proof

I hereby claim:

• I am oremj on github.
• I am oremj (https://keybase.io/oremj) on keybase.
• I have a public key whose fingerprint is 86C6 7AC2 70EB DB77 D8F8 6701 61C0 4E72 F8DE 7B67

To claim this, I am signing this object:

oremj

100000
100000
23860 99865
50763 32492
58198 12402
62324 45154
36169 83580
21219 10561
64338 16841
93963 37330

oremj

{
    "python.linting.enabled": false,
    "python.pythonPath": "/usr/local/bin/python3",
    "window.zoomLevel": -1,
    "editor.formatOnSave": true,
    "vim.neovimPath": "/usr/local/bin/nvim",
    "workbench.list.openMode": "doubleClick",
    "files.associations": {
        "*.jsonnet": "json"
    },

oremj

https://cdn.stubdownloader.services.mozilla.com/builds/firefox-stub/en-US/win/1f023ec81783c9d3cc86c148ab3495f0cd6ae06eb711c05a84fec710ce5377b4/Firefox%20Installer.exe

oremj

# This configuration file for Fluentd is used
# to watch changes to Docker log files that live in the
# directory /var/lib/docker/containers/ and are symbolically
# linked to from the /var/log/containers directory using names that capture the
# pod name and container name. These logs are then submitted to
# Google Cloud Logging which assumes the installation of the cloud-logging plug-in.
#
# Example
# =======
# A line in the Docker log file might look like this JSON:

oremj

'use strict';

const https = require('https');

exports.handler = (event, context, callback) => {
    const request = event.Records[0].cf.request;

    let username = 'Guest';
    if (request.headers['user-name']) {
        username = request.headers['user-name'][0].value;