LLM Wiki CLI
A CLI that creates an Obsidian-ready Markdown vault and installs four skills so an agent maintains the knowledge base.
- 107 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The agent creates and rewrites pages in your folder through the skills
- Source contents go to your agent model
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/jackwener-llm-wikiDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent creates and rewrites pages in your folder through the skills
- Source contents go to your agent model
This is third-party code. Review the repository files before installing.
What it does
LLM Wiki implements Andrej Karpathy's "wiki compiled by an AI" pattern. Instead of re-deriving answers from raw documents, the agent processes sources once into linked Markdown pages and only grows them afterwards. The program itself never calls a model: it provides skill files and a scaffold, and Claude Code, Codex or another agent does the work. The `llm-wiki init` command creates a vault with `wiki-purpose.md`, `wiki-schema.md`, `wiki-log.md`, `wiki/` and `sources/` folders and bootstrap `CLAUDE.md` and `AGENTS.md`, and immediately installs four skills into `.claude/skills/` and `.agents/skills/`: `ingest` turns a source into pages, `query` answers from wiki evidence, `lint` checks base health and `research` brings in external material. Obsidian is the interface for people. After a package upgrade `llm-wiki skill install` refreshes the skills.
Who it is for. Researchers, technical writers and developers who accumulate notes and sources and want an agent to assemble them into a coherent knowledge base.
Good fit when
- Articles and notes pile up and answers have to be found again in raw files.
- An Obsidian vault should grow under agent control following one schema.
- Wiki health is checked periodically: broken links, duplicates, stale pages.
- External research on a topic should immediately become wiki pages.
Not a fit when
- You need a regular RAG system with vector search over raw documents.
- A graphical editor is required: the project deliberately leaves the interface to Obsidian.
- The agent must not be given write access to the notes folder.
Example request
/ingest sources/some-article.md, then /query "What do we know about X?" and /lint across the whole wiki.Limitations
Wiki quality depends entirely on the agent and model running the skills: the program only lays out the scaffold and instructions. The agent rewrites pages, so keep the vault in Git and review changes. Node.js is required and the package is installed globally from npm. Template and structure descriptions are in English. The repository was last updated on 10 July 2026 and uses the Apache-2.0 licence. The README also lists a sync command, so read its documentation before running it.
How to disable. Delete the skills from `.claude/skills/` and `.agents/skills/`, then the package with `npm uninstall -g @jackwener/llm-wiki`. The vault folder and pages stay as regular Markdown files.
FAQ
Does the program call a language model?
No. It only creates the vault and installs skills, and your agent such as Claude Code or Codex calls the model.
What does `llm-wiki init` do?
Creates vault files and bootstrap `CLAUDE.md` and `AGENTS.md`, and installs four skills into `.claude/skills/` and `.agents/skills/`.
Which skills are included?
`ingest` to move a source into the wiki, `query` for answers from pages, `lint` for health checks and `research` for external material.
How does it differ from RAG?
Knowledge is compiled once into linked pages and the agent answers from them instead of re-parsing raw documents on every request.
How to refresh skills after an upgrade?
Run `llm-wiki skill install`: the command refreshes the skill files in the vault.
Discussion
No questions or reviews yet.
Related
CLI toolsAn Obsidian vault template where an agent follows a CLAUDE.md schema to compile sources into a linked wiki and extend it.
LLM Wiki Manager keeps a personal Markdown wiki through an agent: it ingests sources, links pages and answers with citations
CLI for a personal Markdown knowledge base: the agent writes, searches and lints pages with wiki commands; the tool calls no model
Turns session history of Claude Code, Codex, Cursor and Gemini CLI into a searchable wiki on Karpathy's LLM Wiki pattern, locally
A workflow for Claude Code, Codex and Gemini CLI where the agent files source documents into a persistent, linked wiki
A skill set that lets an agent run a personal Obsidian wiki: it compiles knowledge from documents and past sessions and answers with citations
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| LLM Wiki (this card) | no | 107 | |
| LLM Wiki | no | 1,528 | a Claude Code plugin with `/wiki:ingest` and `/wiki:archive` commands; here an npm package creates the vault and skills for two agents |
| Karpathy LLM Wiki | no | 2,732 | one skill for collecting and compiling articles; here a vault and four skills with an init command |
| llmwiki-cli | no | 104 | a CLI for a Markdown base where the agent writes and searches pages by commands; here the focus is skills and an Obsidian scaffold |
README in short
The README describes the idea as compiling knowledge once and querying forever, shows the vault tree and says Obsidian was chosen as the reader interface.