Harpreet Singh

Just tryna figure shit out

[Harpreet Singh @bitvect0r]

Took a walk in Auckland CBD, NZ

[Harpreet Singh @bitvect0r]

Random stroll

[Harpreet Singh @bitvect0r]

Mount Maunganui

[Harpreet Singh @bitvect0r]

Not exactly vibe coding but vibe writing I guess. Just some points from LLM wikis that are getting quite popular.

[Harpreet Singh @bitvect0r]

Reflections from building my LLM wiki. Focus on deeply understanding the idea you’re pursuing, not collecting as much as possible. Build a custom skill tailored to you. Mine uses follow-up questions to explore different angles. Add transcript pulling so you can reflect on videos.

[Harpreet Singh @bitvect0r]

Reflections from building my LLM wiki. Focus on deeply understanding the idea you’re pursuing, not collecting as much as possible. Build a custom skill tailored to you. Mine uses follow-up questions to explore different angles. Add transcript pulling so you can reflect on videos.

[Harpreet Singh @bitvect0r]

@andrewfarah Awesome project! I've bookmarked and forgotten a lot of cool stuff. Could build a cool wiki using this.

I was tryna do something for video content where the LLM pulls the transcript and makes a wiki article.

[Harpreet Singh @bitvect0r]

@aliscodes We can technically use it locally but I found it unusable for most stuff.

I would rather get a much cheaper model like glm and use it with something like openclaude or opencode. Glm5.1 is kinda close to opus.

[Harpreet Singh @bitvect0r]

I use a similar setup as a bookmarking system for the content I wanna revisit. Just forward anything to a Telegram bot, Claude Code fetches the content, summarises and categorises it, and it lands in an Obsidian vault fully text-searchable.

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[Harpreet Singh @bitvect0r]

Vibe coded this RPG game maker. Was a fun project.

[Harpreet Singh @bitvect0r]

Made a simple lightweight RPG engine. Just 1 HTML file and 1 JS file.

There's also a skill so any LLM can use the JS library to build games from a description. Give it a story, a map description, some bosses and make a game.

https://t.co/TDluD3al9f

[Harpreet Singh @bitvect0r]

@kepano I also use this as a bookmarking system. Share anything to a Telegram bot -> Claude Code pulls the content, writes a summary, categorises it -> lands in an Obsidian vault fully text-searchable. Way better than any bookmarking system I've tried.

[Harpreet Singh @bitvect0r]

Made a simple lightweight RPG engine. Just 1 HTML file and 1 JS file.

There's also a skill so any LLM can use the JS library to build games from a description. Give it a story, a map description, some bosses and make a game.

https://t.co/TDluD3al9f

[Harpreet Singh @bitvect0r]

Evening walk

[Harpreet Singh @bitvect0r]

Wanted a workout tracking app, and it was quite easy to get an LLM to build this. It’s just one HTML file, deployed on GitHub Pages and saved to the home screen to be used as an app.

It’s quite fast to add features too, might add workout templates next.

[Harpreet Singh @bitvect0r]

Evening stroll

[Harpreet Singh @bitvect0r]

Snaps from a morning walk in Auckland, NZ

[Harpreet Singh @bitvect0r]

Back when I walked in Glenorchy, New Zealand