LLM Wiki v2
LLM-maintained Markdown wiki where a System One classifier decides where documents and questions go, keeping it fast as it grows.
Overview
LLM Wiki v2 takes your documents, files them into a tidy wiki of Markdown notes and answers questions with citations to the notes it used. It runs locally with any AI provider, including OpenRouter, OpenAI, Gemini, Ollama or any OpenAI-compatible server. A browser UI lets you drop in text, Markdown or PDF files, ask questions, browse a live map of folders, notes and links, and move, describe or delete notes; the files stay as plain Markdown you can keep in git.
By default it runs in CROW mode, where a small System One classifier, Jev via OpenRouter or TypeSafe or a local compatible server, decides where things go: which folder a new document belongs in and which note a question should be routed to, guided by folder descriptions that librarian and researcher agents maintain. Without a classifier provider it falls back to classic mode, where the LLM decides by itself. A demo wiki works without a key. The listing notes it is built on Google's Open Knowledge Format and licensed AGPL-3.0.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which folder should this new document be filed in?(inferred) | — |
| Choice | Which note or folder should answer this question? | — |
How it works
- Drop documents into the wiki
- Classifier routes each to a folder; the LLM writes or merges notes
- Ask a question; classifier routes it to relevant notes
- LLM answers with citations
- Classic mode is used if no classifier is configured
Strengths
- Stays fast and cheap as the wiki grows
- Plain Markdown files you own
- Works with free local models via Ollama
- Demo needs no key
Caveats
- Answer quality depends on the chosen LLM
- Classifier needs a separate provider
- AGPL-3.0 licence
Getting started
uv tool install "llmw2[server] @ git+https://github.com/crow-hq/llm-wiki-v2" then llmwiki2.At a glance
- Type
- Service
- Status
- beta
- Language
- Python
- Runtime
- local web app
- Has code
- yes
- License
- AGPL-3.0
Use cases and integrations
How curated lists describe it
An LLM-maintained wiki on Google's Open Knowledge Format with librarian and researcher agents, where a System One classifier (Jev via OpenRouter or TypeSafe, or a local server) routes each question to the right note. AGPL-3.0.
MrJev/awesome-jev
Frequently asked questions
What does the classifier decide?
Where documents are filed and which notes a question is routed to.
Can I run it without cloud keys?
Yes, with Ollama for the LLM; the classifier can use a local server or be skipped.
Where are my notes stored?
As Markdown in ~/llm-wiki.
Related projects
Privacy-first personal knowledge management system, listed as integrating Jev for document classification and tag routing.
PluginproductionTypeScriptGo terminal app that files documents using five Jev questions: category, sensitivity, urgency, whether action is needed and action state.
ProjectbetaGoObsidian plugin that ranks existing vault folders and tags for the active note with Jev and only moves or tags after you confirm.
ProjectbetaTypeScriptDocker service that classifies the Paperless-ngx inbox with Jev (or local Clef via Ollama), setting confident fields and queueing the rest.
ProjectbetamacOS menu-bar link router that opens each link in the right browser profile, asking Jev when no local rule matches and learning rules.
ProjectbetaSwiftCLI that classifies documents with a Jev Choice and moves them into category folders, with a pluggable backend for local models.
ProjectbetaPython