Browse / LLM wiki / Second brain / second-brain
second-brain
ZenekeZene
AI-maintained personal wiki.
Open ZenekeZene/second-brain → Checked against the four rules.
Built for Claude Code
ships .claude CLAUDE.md and nothing for other agents.
In its own words
Raw material in, interconnected markdown articles out. Telegram bot, X sync, and scheduled agents for daily compilation, weekly linting, and monthly health checks.
A self-hosted AI that reads what you save, builds your knowledge base overnight, and briefs you every morning. Runs on a $35 Raspberry Pi.
Save a link from your phone at midnight. Wake up to a wiki article, cross-linked with everything you already know, with a morning briefing waiting in Telegram.
Not a RAG. Not a chatbot with memory. Not a tool you have to open. An autonomous system that compounds your knowledge while you sleep.
How it feels
No laptop open. No command to run. No maintenance.
Quick Start
Requirements: Claude Code CLI · Node.js ≥ 20
Open Claude Code inside the directory:
That's it. Your first wiki article will be in wiki/.
How it works
Raw material goes into raw/, an LLM compiles it into interconnected articles in wiki/, and queries feed new insights back into the wiki.
Installation
Prerequisites
Two compilation backends are available — you need at least one. See Compile Backend for details and how to switch between them from the wiki UI.
Usage
From Claude Code (conversational mode)
Open Claude Code inside this directory:
CLAUDE.md tells the LLM how the system works. The brain: prefix is a chat convention — type it as a regular message to Claude Code. It is not a terminal command.
Compile pending items
Claude reads all pending items, integrates them into interconnected wiki articles, and updates INDEX.md.
The repository's own description, then 9% of its README. Read the rest on GitHub ↗
Read from the repository's git tree on 2026-08-17.