add-karpathy-llm-wiki

Ingest sources into a persistent cross-referenced wiki within NinjaClaw groups.

3|Updated Apr 19, 2026
One-click install
npx skills add https://github.com/OfirGavish/NinjaClaw-Nano --skill add-karpathy-llm-wiki-ofirgavish
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: add-karpathy-llm-wiki
Source: https://github.com/OfirGavish/NinjaClaw-Nano/tree/main/.copilot/skills/add-karpathy-llm-wiki
Command: npx skills add https://github.com/OfirGavish/NinjaClaw-Nano --skill add-karpathy-llm-wiki-ofirgavish

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Establishes a process to create and maintain a persistent, cross-referenced wiki knowledge base within NinjaClaw groups using Karpathy's LLM Wiki pattern, so knowledge accumulates and evolves instead of being rederived per query.

Core Features & Use Cases

  • Three-layer architecture (raw sources, the wiki, and the schema) that ingests sources, builds linked wiki pages, and enforces conventions.
  • Step-by-step group setup: design collaboration, per-file ingestion discipline, and ongoing maintenance through ingest, query, and lint operations.
  • Use cases include personal knowledge management, team documentation, and research mentoring within a NinjaClaw group.

Quick Start

Clone the Karpathy LLM Wiki pattern, configure your group, and start ingesting your first source to build the wiki.

Frequently Asked Questions about add-karpathy-llm-wiki

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a persistent wiki knowledge base for my team?

To build a persistent wiki knowledge base, this Skill ingests sources step-by-step into a NinjaClaw group, extracting entities and cross-referencing pages so knowledge accumulates instead of being rederived per query.

What is the Karpathy LLM wiki pattern for knowledge management?

The Karpathy LLM wiki pattern uses a three-layer architecture: raw sources, the wiki pages, and an enforced schema. It processes sources individually to build linked pages and maintain a collaborative knowledge base.

How do I configure a NinjaClaw group for stepwise knowledge ingestion?

Configuring stepwise knowledge ingestion requires setting up a container-based workflow with a defined schema in a COPILOT.md file, then applying explicit per-file ingest discipline to ensure deep integration and avoid batch-processing.

Can I use batch processing to ingest multiple files into the wiki at once?

No, you should not use batch processing for wiki ingestion. The pattern requires explicit per-file ingest discipline to avoid batch-processing and ensure deep, cross-referenced integration of each source.

What is the best way to maintain a collaborative wiki across multiple groups?

The best way to maintain a collaborative wiki is ongoing maintenance through structured ingest, query, and lint operations within your NinjaClaw groups, ensuring knowledge stays cross-referenced and evolves over time.