llm-wiki

A reusable template for building and maintaining AI-powered personal knowledge bases with Claude Code. Drop your source documents, get a structured, interlinked wiki that compounds over time.

Architecture: wiki data lives in this git repo (projects/ /wiki/). Raw source files (PDFs, docs) stay on Google Drive and are pulled on demand. Everything is saved with git commit + git push.

The idea

Most LLM-document tools retrieve from raw sources at query time — rediscovering knowledge from scratch on every question. Nothing accumulates.

This template does something different: Claude builds and maintains a persistent wiki between you and your raw sources. When you add a new document, Claude doesn't just index it — it reads it, extracts key information, and integrates it into the existing wiki: updating entity pages, flagging contradictions, strengthening the evolving synthesis.

The wiki is a compounding artifact. The cross-references are already there. The contradictions have already been flagged. Every source you add makes it richer.

Two roles, two models

Claude builds rich, structured wiki pages optimized so that Groq can navigate them efficiently. Groq never needs to see the entire wiki — it reads a lightweight index to find the right pages, then searches their full text.

Quickstart

1. Clone and open in VS Code

Or open as a GitHub Codespace — the devcontainer sets everything up automatically.

3. Create a project

Add a row to projects.md: