Skip to content

Instantly share code, notes, and snippets.

View fkztw's full-sized avatar
📺
Hello World!

Frank Zheng fkztw

📺
Hello World!
350 followers · 377 following
  • Taiwan
  • 11:35 (UTC +08:00)
View GitHub Profile
@fkztw
fkztw / satoshistreasure.md
Created April 17, 2019 03:02 — forked from johncantrell97/satoshistreasure.md
How I Obtained Satoshi's Treasure Keys 1, 2, and 3 in Minutes

Today (April 16th 2019 at noon) the first major clues to discover key #1 was set to be released in a few cities. A QR code with the words 'orbital' were found at these locations and looked like this: (https://imgur.com/a/6rNmz7T). If you read the QR code with your phone you will be directed to this url: https://satoshistreasure.xyz/k1

At this URL you are prompted to input a passphrase to decrypt the first shard. An obvious first guess was to try the word 'orbital' from the QR code. Not suprisingly this worked! This reveals a congratulations page and presents the first key shard:

ST-0001-a36e904f9431ff6b18079881a20af2b3403b86b4a6bace5f3a6a47e945b95cce937c415bedaad6c86bb86b59f0b1d137442537a8.

Now, we were supposed to wait until April 17th to get clues from the other cities for keys #2 and #3 but that wouldn't stop me from digging around with all the new information we had. All that time "playing" notpron (http://notpron.org/notpron/) years ago was going to help me here.

The first thing I noticed was

@fkztw
fkztw / llm-wiki.md
Created April 23, 2026 09:26 — forked from karpathy/llm-wiki.md
llm-wiki

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.