Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
Back to list

Add Karpathy LLM Wiki (NanoClaw)

A guided, re-run-safe setup that wires Karpathy's LLM Wiki pattern into a NanoClaw group as a persistent knowledge base.

AutomationIntermediate30,57312,843AI score 8/10Last updated: Aug 20, 2026

What it does

  • Implements Andrej Karpathy's LLM Wiki pattern (three layers: sources / wiki / schema; three operations: ingest · query · lint) inside a NanoClaw group.
  • Picks the target group (existing main chat or a new dedicated wiki group), then creates wiki/ and sources/ plus starter index.md and log.md.
  • Generates a tailored container/skills/wiki/SKILL.md (the schema layer) and inserts/replaces a Wiki section in the group's CLAUDE.md using marker comments.
  • Checks whether the agent can handle your planned source formats (PDF, images, voice) and helps install the missing capability skills.
  • Optionally schedules weekly/monthly lint health checks via the schedule_task MCP tool, then walks you through restarting the service.
  • Every step is idempotent: existing wiki files and a customized wiki skill are never silently clobbered.

Who it's for

  • People actively running NanoClaw who want a chat group to double as a growing knowledge base.
  • Anyone who wants a personal wiki where dropping a link, PDF, or voice note yields structured notes.
  • Readers of Karpathy's LLM Wiki write-up who want a concrete implementation.

Examples

  1. Say "add wiki" → the skill summarizes the pattern, spins up a dedicated "paper reading" group, scaffolds wiki/, sources/, index.md, log.md, and injects the CLAUDE.md wiki section.
  2. Point it at a folder of 12 PDFs → it processes them one at a time: read, discuss takeaways, write summary/entity/concept pages, update index and log, then move on.
  3. Send an article URL → instead of a WebFetch summary it runs curl -sLo sources/... "<url>" to capture full text for deep ingestion.
  4. Choose "weekly health checks" → a recurring lint task is registered through schedule_task.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/nanocoai/nanoclaw/HEAD/.claude/skills/add-karpathy-llm-wiki/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the .claude/skills/add-karpathy-llm-wiki folder from the GitHub repo nanocoai/nanoclaw into my ~/.claude/skills/add-karpathy-llm-wiki/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/nanocoai/nanoclaw.git && mkdir -p ~/.claude/skills && cp -r nanoclaw/.claude/skills/add-karpathy-llm-wiki ~/.claude/skills/

This is a third-party skill. Check the source repository before installing.

  1. Make sure you already have a working NanoClaw install (pnpm plus the setup/ scripts).
  2. Clone the repo: git clone https://github.com/nanocoai/nanoclaw.git
  3. Copy the skill: mkdir -p ~/.claude/skills && cp -r nanoclaw/.claude/skills/add-karpathy-llm-wiki ~/.claude/skills/
    • Or leave it in the NanoClaw project's .claude/skills/ and launch Claude Code from the project root.
  4. Confirm the folder contains both SKILL.md and llm-wiki.md.
  5. Open Claude Code at your NanoClaw project root and type "add wiki".
  6. Answer the prompts: which group, the wiki's domain, and which source types you'll ingest.
  7. Run the final restart command (macOS launchctl kickstart, Linux systemctl --user restart), then send a test source to the wiki group.