A persistent knowledge management system that compiles raw information into a structured, interlinked wiki based on the Karpathy LLM-Wiki methodology.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install llm-wiki-admin
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 llm-wiki-admin 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 LLM-Wiki Knowledge Manager?
The LLM-Wiki skill is a sophisticated tool for Openclaw Skills designed to move beyond simple RAG (Retrieval-Augmented Generation) toward knowledge compilation. Instead of re-deriving answers from scratch for every query, this skill synthesizes raw data into a persistent Markdown-based wiki. This approach ensures that insights, cross-references, and syntheses are captured permanently, allowing your knowledge base to compound in value over time.
Rooted in the Karpathy LLM-Wiki philosophy, the system operates on a three-layer architecture: immutable raw data, AI-maintained wiki pages, and a set of 'constitutional' documents (schema and purpose) that guide the AI's behavior. By acting as a professional knowledge manager rather than a generic assistant, it ensures that your Obsidian vault becomes a high-fidelity second brain.
LLM-Wiki Knowledge Manager Use Cases
- Compiling complex research papers, books, and articles into a structured, interlinked knowledge repository.
- Managing team meeting records and project decisions to ensure a single, AI-curated source of truth.
- Building a personal growth system that tracks goals, habits, and reflections with cross-referenced insights.
- Transforming fragmented reading notes into a cohesive wiki with identified themes and character threads.
How LLM-Wiki Knowledge Manager Works
- Initialization: The skill interviews the user to establish the theme, language, and specific use case, generating the core architecture files.
- Ingestion: Raw documents provided by the user are analyzed for key entities, concepts, and relationships.
- Compilation: The AI updates existing wiki pages or creates new ones, integrating the new data while preserving granular details and identifying contradictions.
- Indexing and Logging: Every change is tracked in a reverse-chronological log and reflected in a central index and high-level overview.
- Querying: Users can ask questions directly to the wiki, and the AI provides answers with internal citations and links back to the compiled pages.
- Health Maintenance: The system scans for orphaned pages, missing links, or outdated content to keep the vault clean.
LLM-Wiki Knowledge Manager Setup
To get started with this skill in your Openclaw Skills environment, ensure you have the prerequisite Obsidian skill installed. Follow these steps:
- Trigger the initialization process by asking the agent to 'initialize a new wiki'.
- Provide the local path to your Obsidian Vault when prompted.
- Complete the brief interview regarding your subject area and language preferences.
- The skill will automatically generate the following directory structure:
mkdir raw
mkdir wiki
touch schema.md
touch purpose.md
- Place your source documents into the
raw/folder to begin the ingestion process.
LLM-Wiki Knowledge Manager Data Schema & Taxonomy
The LLM-Wiki skill organizes information into a strict hierarchy to ensure technical accuracy and discoverability within Openclaw Skills:
| Component | Description |
|---|---|
raw/ |
Immutable directory for original source files (PDFs, Markdown, text). |
wiki/ |
The compiled knowledge base containing AI-generated Markdown pages. |
schema.md |
The 'Constitution' defining page types, naming conventions, and metadata rules. |
purpose.md |
The 'North Star' document defining the project goals and boundaries. |
wiki/index.md |
A structured directory of all compiled pages. |
wiki/log.md |
A reverse-chronological record of every ingestion, query, and linting action. |
LLM-Wiki Knowledge Manager Advanced Features
- Conflict Detection: Instead of overwriting data, the skill identifies and highlights contradictions between different sources using specialized warning callouts.
- Bi-Directional Linking: Automatically generates Obsidian-compatible wikilinks for effortless navigation within your local IDE.
- Scenario-Specific Templates: Supports specialized schemas for Research, Reading, Business, and Personal Growth out of the box.
- Integrity Linting: Automated health checks to detect orphaned pages, broken wikilinks, or missing frontmatter metadata.
- Granular Data Preservation: Strict rules ensure that specific numbers, percentages, and original story beats are never lost during the summarization process.
SKILL.md
Loading
Related Openclaw Skills
Style Design Generator
A sophisticated aesthetic generation engine designed to create, mix, and adapt 100 unique style schemes across multiple creative domains.
AI Video Notes
An AI-powered tool that converts public video URLs into structured document, outline, or graphic-text notes via the SkillBoss API Hub.
fitbuddy
A comprehensive fitness assistant for tracking nutrition, weight, and exercise with automated metabolic calculations and safety-first health logic.
Anti-Rug Web3 Token Security Scanner
A professional-grade Web3 security scanner that uses a cross-validation engine to detect rug pulls, honeypots, and contract risks across multiple EVM chains.
Summarize Skill
A lightning-fast CLI utility for distilling URLs, local files, and YouTube content into concise summaries using leading AI models.
Larry Page Distillation Master
A high-level strategic persona that distills Larry Page's 10X thinking and AI-native principles to drive moonshot innovation and product excellence.
Featured*
