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.

twinforces

Grumpy Analysis (topic deep-dive)

Invocation (paste into any tool that can fetch URLs):

Read this gist and apply it fully:
https://gist.github.com/twinforces/534d9b662de4a010c1c4ebad934cd99a
Topic: [TOPIC]
Optional: focus on [angle / hearing / bill / claim]; depth [brief | standard | deep]

Shorthand once the gist is known:

rodrigoborgesdeoliveira

• Akan:
  • Nyankómàgó
  • Mǎnu
  • Meńsã́
  • Nsĩã́
  • Dúkũ
  • Ɔkwán
  • Nyamékyε
  • Obím̀pέ

jordan-sussman

name	auto-review-loop
description	Use when the user types /auto-review-loop, or asks to start, restart or run the recurring loop that reviews the PRs in a repo that request their review and auto-approves the good ones. Also use when they ask to post the comments for a PR that loop rejected.

PR auto-review loop

Reviews every open, non-draft PR in one GitHub repo that requests the user's review. Bare-approves the good ones. Hands rejects to the user. Nothing else reaches GitHub unless the user asks for it in their own words.

Requires a /review skill that prints an approve or reject verdict. The finding-verifier agent is optional: the review skill may use it to check findings.

genadyp

link

#!/bin/bash

#File: tree-md

tree=$(tree -tf --noreport -I '*~' --charset ascii $1 |

bmaupin

See also:

• Free backend hosting
• https://free-for.dev/#/?id=dbaas

Service	Type	Storage	Limitations
Amazon DynamoDB	⚠️ Proprietary NoSQL	25 GB	⚠️ Payment method required
Amazon RDS	⚠️ Proprietary RDBMS		⚠️ Only free for 1 year
Azure SQL Database	MS SQL Server		⚠️ Only free for 1 year
👉 Clever Cloud	PostgreSQL, MySQL, MongoDB, Redis	256 MB (PostgreSQL)	Max 5 connections (PostgreSQL)

Balzabu

Vulnerable Test Websites

Domain	URL	Environment / Project	Technologies	Vulnerabilities / Scenarios
public-firing-range.appspot.com	https://public-firing-range.appspot.com/	Google Firing Range	Google App Engine	Address XSS, Angular XSS, bad JavaScript imports, clickjacking, CORS issues, DOM XSS, escaped XSS, Flash injection, leaked HttpOnly cookie, mixed content, redirect XSS, reflected XSS, remote inclusion XSS, reverse clickjacking, HSTS tests, tag-based XSS, URL-based DOM XSS, vulnerable libraries, insecure third-party scripts
pentest-ground.com:81	https://pentest-ground.com:81/	GuardianLeaks	Web App	XSS, SSRF, Code Injection
pentest-ground.com:4280	https://pentest-ground.com:4280/	Damn Vulnerable Web Application	Classic Web App	CSRF, XSS, SQLi
pentest-ground.com:5013	https://pentest-ground.com:5013/	Damn Vulnerable GraphQL Application	GraphQL	Command Injection, XSS, SQLi
pentest-ground.com:7001	https://pentest-ground.c

NathanWalker

Electron: a non-activating, keyboard-focusable panel on macOS

For electron/electron#52915: a Spotlight-like input panel that takes typing while the app the user was in stays active.

NativeScript's macOS runtime makes every Objective-C class a global in Electron's main process, and JavaScript classes can subclass them. Nothing to compile, no addon, no patch to Electron.

See it running

On an Apple Silicon Mac with macOS 14.4 or later: