Browse / LLM wiki / Second brain / personal-llm-wiki

personal-llm-wiki

Kengston

🧠 Personal AI second brain — a self-maintaining Markdown wiki (Karpathy LLM-wiki pattern).

Open Kengston/personal-llm-wiki → Checked against the four rules.

agentic-aiai-assistantanthropicclaudeclaude-codefastifyknowledge-baseknowledge-graphllmllm-agentmarkdownnodejs

Works with any agent

ships AGENTS.md, the cross-tool instruction file.

In its own words

AboutKengston/personal-llm-wiki

TypeScript, no vector DB, Claude-native engine, Telegram bridge.

README.md

personal-llm-wiki — «Второй мозг»

A personal AI second brain — a self-maintaining Markdown wiki on Andrej Karpathy's LLM-wiki pattern. No vector DB · no embedder · Claude-native · Telegram (reactive + proactive) · two-repo privacy

A personal AI assistant built as a persistent, interlinked Markdown wiki that an LLM agent maintains incrementally — Andrej Karpathy's LLM-wiki pattern. No vector DB. No embedder. A subscription-CLI agent (Claude Code, the official claude binary) does the semantic ranking by reading the wiki, runs across three execution layers (reactive · scheduled · event), a thin Telegram bridge makes it reactive and proactive, and a hard two-repo split keeps the framework public while your data stays private. The engine sits behind a portable seam — Claude-native today, Grok/Codex are deferred adapter slots.

The repository's own description, then 6% of its README. Read the rest on GitHub ↗

Read from the repository's git tree on 2026-08-17.