An AI-driven compilation-over-retrieval wiki system that transforms raw sources into a structured, compounding knowledge base for agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install wiki-system
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install wiki-system using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Karpathy Wiki?
This skill implements the compilation-over-retrieval pattern popularized by Andrej Karpathy. Unlike traditional RAG systems that retrieve and forget, this Openclaw Skills extension builds a durable markdown wiki. It allows AI agents to synthesize information from various sources into cross-referenced pages that grow more valuable over time. By shifting the burden of bookkeeping to the agent, users can focus on high-level curation while the AI handles the heavy lifting of summarization and organization.
Karpathy Wiki Use Cases
- Automating the synthesis of academic papers, technical articles, and PDFs into a centralized knowledge hub.
- Maintaining a cross-referenced personal or professional wiki without manual formatting or maintenance overhead.
- Preventing insight loss from chat sessions by compounding external data into durable, long-term memory.
- Performing automated weekly health checks to identify broken links, orphans, or contradictions in research documents.
How Karpathy Wiki Works
- Users place raw source files like PDFs, papers, or URLs into a designated wiki-sources directory.
- The ingestion process extracts core claims and creates or updates structured markdown pages with YAML frontmatter.
- The agent automatically adds wikilinks to cross-reference related topics across the entire database, ensuring knowledge connectivity.
- OpenClaw memory searches index these pages, making the compiled knowledge available for synthesis in future queries.
- A weekly linting process runs to ensure the wiki remains clean, organized, and free of metadata errors or stale links.
Karpathy Wiki Setup
To get started with this skill for Openclaw Skills, run the basic setup script to initialize directories, templates, and memory references:
bash skills/karpathy-wiki/scripts/wiki-setup.sh
For a full installation including automated cron jobs for ingestion and cloud synchronization via SuperMemory, use the following command:
bash skills/karpathy-wiki/scripts/wiki-setup.sh --all --tz America/New_York
Karpathy Wiki Data Schema & Taxonomy
The skill organizes data into a structured hierarchy within the workspace to ensure high searchability and compatibility:
| Component | Path | Description |
|---|---|---|
| Wiki Pages | memory/wiki/*.md |
Structured markdown files using domain-prefixed kebab-case names and YAML frontmatter. |
| Raw Sources | work/wiki-sources/ |
Immutable input files including PDFs, papers, and articles. |
| Catalog | memory/wiki/index.md |
A structured page catalog that helps the agent navigate existing topics. |
| Rules | WIKI-SCHEMA.md |
Defines the rules for structure, naming conventions, and domain categorization. |
Karpathy Wiki Advanced Features
- Automated Cron Jobs: Schedule daily ingestion (4AM) and weekly linting (Sunday 3:30AM) for zero-touch maintenance.
- Obsidian Compatibility: The use of standard wikilink syntax allows you to browse and visualize your knowledge graph directly in Obsidian.
- Dreaming Integration: Coexists with the Dreaming skill, allowing conversation signals and external knowledge to grow in parallel without file conflicts.
- Multi-Backend Support: Native compatibility with QMD for semantic search and Honcho for cross-session user modeling.
- Cloud Persistence: Optional sync scripts to maintain your wiki across multiple devices using SuperMemory.
SKILL.md
Loading
Related Openclaw Skills
Text Cleaner Lite
A streamlined utility for normalizing whitespace, removing redundant blank lines, and trimming raw text inputs.
ClawHub Publish Skill
A specialized automation tool for validating and deploying AI agent skills directly to the ClawHub registry.
bc-calc
Perform arbitrary-precision arithmetic and complex mathematical computations using the Unix bc engine.
Text Cleaner CLI
A specialized command-line utility for removing redundant whitespace and empty lines from text and JSON files.
Vertical Home Renovation Estimator
An AI-powered estimation tool that generates structured home renovation cost reports based on area, quality grade, and location.
Hoist
Hoist is an AI-native CLI designed to help agents deploy and manage apps, servers, and databases on VPS providers with automated JSON output.
Featured*
