rohitg00

LLM Wiki v2

A pattern for building personal knowledge bases using LLMs. Extended with lessons from building agentmemory 20K+ Stars ⭐️, a persistent memory engine for AI coding agents.

This builds on Andrej Karpathy's original LLM Wiki idea file. Everything in the original still applies. This document adds what we learned running the pattern in production: what breaks at scale, what's missing, and what separates a wiki that stays useful from one that rots.

What the original gets right

The core insight is correct: stop re-deriving, start compiling. RAG retrieves and forgets. A wiki accumulates and compounds. The three-layer architecture (raw sources, wiki, schema) works. The operations (ingest, query, lint) cover the basics. If you haven't read the original, start there.

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.

ditzel

using System;
using System.Collections;
using System.Collections.Generic;
using UnityEngine;

public class MeshDestroy : MonoBehaviour
{
    private bool edgeSet = false;
    private Vector3 edgeVertex = Vector3.zero;
    private Vector2 edgeUV = Vector2.zero;

ditzel

/*

Set this on an empty game object positioned at (0,0,0) and attach your active camera.
The script only runs on mobile devices or the remote app.

*/

using UnityEngine;

class ScrollAndPinch : MonoBehaviour

Dark4Codrutz

Windows Registry Editor Version 5.00
;Reveals the most used hidden power settings in Windows 10. The value Attributes=2 means reveal, where value of 1 or 0 means hide from Power Options.
;Good Tweaking!

;WLAN
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\19cbb8fa-5279-450e-9fac-8a3d5fedd0c1\12bbebe6-58d6-4636-95bb-3217ef867c1a]
"Attributes"=dword:00000002

;Power Scheme Personality
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\245d8541-3943-4422-b025-13a784f679b7]

ditzel

using UnityEngine;

namespace Ditzelgames
{
    public static class PhysicsHelper
    {

        public static void ApplyForceToReachVelocity(Rigidbody rigidbody, Vector3 velocity, float force = 1, ForceMode mode = ForceMode.Force)
        {
            if (force == 0 || velocity.magnitude == 0)

ditzel

using System;
using System.Collections;
using System.Collections.Generic;
using UnityEngine;
using UnityEngine.Profiling;

public class KdTree<T> : IEnumerable<T>, IEnumerable where T : Component
{
    protected KdNode _root;
    protected KdNode _last;

reunono

using UnityEngine;
using UnityEngine.UI;
using System.Collections;
using System.Collections.Generic;

/// <summary>
/// Add this component to your Cinemachine Virtual Camera to have it shake when calling its ShakeCamera methods.
/// </summary>
public class CinemachineCameraShaker : MonoBehaviour
{

unity3dcollege

using UnityEngine;

public class TerrainDetector
{
    private TerrainData terrainData;
    private int alphamapWidth;
    private int alphamapHeight;
    private float[,,] splatmapData;
    private int numTextures;

runewake2

using System.Collections;
using System.Collections.Generic;
using System.Linq;
using UnityEngine;

[RequireComponent(typeof(GeneratePlaneMesh))]
public class DeformableMesh : MonoBehaviour {

    public float maximumDepression;
    public List<Vector3> originalVertices;