Andrej Karpathy's LLM Wiki concept

Your notes.
A living wiki.

Write in Obsidian. Every note connects to a network that keeps growing.

Get the Plugin Get Obsidian
Obsidian required.
Scroll
Why It Matters

Nothing you read should go to waste.

Lose it, or grow it?

Before — Saving

Save an article. Forget where. Search later. Nothing.

After — Saving

Drop it in sources/. Ideas found, pages built, summary ready. In seconds.

Before — Connecting

Link two notes by hand. One changes. The other stays old. The link dies.

After — Connecting

Links stay alive. When new information contradicts the old, every related page stays current.

Before — Growing

A year of notes. Just files. Every project starts over. Same reading. Same conclusions.

After — Growing

A year builds a wiki. Every new question draws on everything you've collected. Your knowledge grows.

How It Works

From note to wiki

One note. A world of connections.

Source
sources/sleep-hygiene.md
1# Why We Sleep: Key Findings
2
3Sleep is the single most effective thing we can do to reset our brain and body.
4
5## Sleep Architecture
6- NREM Stage 3 (deep sleep): physical restoration, immune function
7- REM: emotional processing, creativity, memory consolidation
8- Each cycle ~90 min, 4-5 cycles per night ideal
9
10## The Cost of Sleep Debt
11- After 1 week of 6h/night: cognitive performance equals 0.05% BAC
12- Chronic sleep debt linked to Alzheimer's — glymphatic clearance drops 60%
13- 5h sleep → testosterone drops 10–15% in healthy young men
14
15## Practical Sleep Hygiene
16- Optimal bedroom temperature: 18–19°C
17- No blue light 90 min before bed — melanopsin sensitivity peaks at 480nm
18- Caffeine half-life 6h: last intake before 2pm
19- Wake-time consistency matters more than bedtime consistency
20
21## Open Questions
22- Do naps compensate for night debt? Evidence mixed
23- Melatonin: 0.3mg as effective as 3mg. Most supplements are overdosed.
154 words · 5 sections · 23 lines
Features

What changes for you

Not features. A new relationship with everything you read.

Zero effort

It just happens

Drop it in sources/. AI reads, extracts, builds. No folders. No tags. No work.

Natural

Just talk

Ask like you would a friend. Watch it reason, then answer with sources you can open and trust.

Connected

Living links

Every page joins your Obsidian graph. Open Graph View — and watch your knowledge grow.

On your terms

Stays current

New files? Auto-ingested. Contradictions? Found. Health? Checked. On your terms.

And more:
Every change, traceable One click fixes everything What you finalize, stays Every answer, traceable Your tags. Your call. Fine to coarse. You decide. 10-language outputSee it reasonCatches contradictionsFinds duplicatesRecovers from hiccupsSkips what's done
Get Started

Ready in minutes

Obsidian is all you need.

01

Install

In Obsidian: Settings → Community plugins → Browse, search Karpathy LLM Wiki, click Install then Enable.

Open Plugin Page
02

Configure

Settings → Karpathy LLM Wiki. Pick your LLM, enter key, test, save.

03

Use

Cmd+P (or Ctrl+P). Type Ingest to add sources. Type Query to ask.

FAQ

Questions, answered

Everything worth knowing, before you start.

What does this plugin actually do?

Drop any note into your Obsidian vault. The AI extracts people, concepts, and theories, then auto-generates interlinked wiki pages with bidirectional links. Ask "what did I write about X?" — the answer comes from your own notes, not the internet.

What are the minimum requirements?

Obsidian v1.11.0+ (desktop: Windows, macOS, Linux) and an LLM provider API key. Supported: DeepSeek, Gemini, Claude, GPT, Kimi, GLM, MiniMax, OpenRouter, or any custom endpoint. No API key needed for local models via Ollama or LM Studio.

Which model should I choose? How much will it cost?

Long-context models recommended — they process your entire wiki in one pass. A single ingest costs $0.05–$0.50. All costs go to the LLM provider; the plugin itself is free and open source.

Can I run it fully offline with local models?

Yes. Install Ollama or LM Studio, pull a model, select it as your provider — no API key, nothing leaves your machine. Cloud providers are better for heavy ingestion (larger context); local models work great for everyday query.

How do I get help or report bugs?

Use GitHub Issues for bug reports, GitHub Discussions for questions. Also find the plugin on the Obsidian Community Plugin page and leave a review.

Ecosystem

Every plugin, amplified

Not an island. Every tool you love, now better.

Graph View

See links

+ Karpathy LLM Wiki

Wiki gives your graph meaning. Hub pages, clusters, orphans. Exploration, not decoration.

Web Clipper

Save articles

+ Karpathy LLM Wiki

Drop in sources/. AI extracts, links, updates. One clip becomes 10+ wiki pages.

Dataview

Query data

+ Karpathy LLM Wiki

AI adds structure. Tags, dates, categories. Empty tables become living dashboards.

Git

Track versions

+ Karpathy LLM Wiki

Watch your knowledge evolve. Every commit tells the story of how you grew.

Marp

Make slides

+ Karpathy LLM Wiki

Ask: "Summarize my research for a 10-minute talk." Wiki builds slides from your knowledge.

Canvas

Visual canvas

+ Karpathy LLM Wiki

AI builds concept maps, timelines, decision trees from your knowledge. No more blank canvas.

Just the beginning. When your knowledge has structure, every plugin becomes more powerful.
Bring Your Own AI

Your model. Your rules.

Twelve providers, cloud or local. Switch anytime — lock into nothing.

Switch any time

Twelve providers, one dropdown apart. Outgrow one, switch to the next — your wiki never notices.

Privacy first

No backend. No tracking. Nothing stored. Your notes only travel to the AI you choose — or stay fully offline with a local model.

Costs almost nothing

The plugin is free. A full ingest runs you $0.05–$0.50 on the LLM side — pick a value model and you'll barely notice it.

Open source

MIT licensed, every line on GitHub. Fork it, audit it, make it yours.

→ GitHub
Twelve providers, ready to go

Cloud or local — use the one you trust.

DeepSeek
Best value
Best value
Anthropic
Claude
Google
Gemini
OpenAI
GPT
Kimi
Moonshot
GLM
Zhipu AI
MiniMax
Cloud
OpenRouter
Any model
Ollama
Local
Local
LM Studio
Local
Local
Anthropic-Compatible
Custom endpoint
OpenAI-Compatible
Custom endpoint

Long-context models read your whole wiki in a single pass — the complete picture, not fragmented snippets.

Ready to build your wiki?

Open source. Community driven. Just Obsidian.

Add to Obsidian