yi-john-huang

Loreloom — AI skill for Claude Code

AI community

An AI-first Obsidian vault framework that weaves sources into living knowledge.

How to install Loreloom

This entry records only its repository, not the path inside it, so there is no exact command to give. Open yi-john-huang/loreloom and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Loreloom does

An AI-first Obsidian vault framework that weaves sources into living knowledge.

Alternatives in AI

Claudian — An Obsidian plugin that embeds Claude Code as an AI collaborator in your vault 7.7k ★
Karpathy LLM Wiki — Agent Skills-compatible LLM wiki for Claude Code, Cursor, and Codex 618 ★
Obsidian + Claude Code: Rebuild Your Second Brain With AI — Build an AI-powered personal knowledge management system from scratch 540 ★

README

Loreloom

An AI-first Obsidian vault framework that weaves captured sources into reviewed knowledge, generated wiki pages, and a connected multi-topic graph.

The framework treats Markdown as the durable source format and separates captured material, reviewed knowledge, and generated synthesis. It is intentionally model-agnostic: your vault should outlive any particular AI provider.

[!IMPORTANT] This repository contains example content only. Create a new private repository from this template before adding personal, confidential, or copyrighted material.

Core model

Owner Asset / absolute HTTP(S) URL / Inbox / Daily
                         |
                         v
Source processing (editable intake)
                         |
                         v
Owner review (direct or signed source_review)
                         |
                         v
Current-byte Asset revalidation
                         |
                         v
Concept draft -> owner claim/citation review -> optional evergreen
                         |
                         +------------------+
                         v                  v
                       Wiki               MOCs
                         |
                         v
                   Projects / Areas

One vault, many topics. Technology, food, travel, career, and other subjects connect through links and metadata instead of separate vaults.
Evidence before synthesis. Source material is preserved; durable claims point back to it.
Human-reviewed truth. Knowledge/ is canonical. AI may draft there, but a human changes status to evergreen.
Generated pages are replaceable. Wiki/ pages declare their inputs and can be rebuilt when models or prompts improve.
Agents are governed. AGENTS.md and .agents/policies/ define what an agent may read, create, or change.
One root by default. The active Codex session runs bounded skills sequentially; custom read-on