Lore — Your AI-Powered Second Brain

Lore is an open-source second brain that uses Claude to learn on your behalf, organise what you know, and quiz you on it — all in plain Markdown files you own.

→ Website & Docs · Portal

---

How It Works

1. You set goals in brain/self/goals.md
2. Run /learn → Claude finds free resources aligned to your goals and writes wiki articles
3. Run /ask what should I focus on this week? → Claude answers from your wiki
4. Run /assess → interactive knowledge quiz on what you've learned
5. Run /reflect → daily journal entry consolidating decisions and signals

Everything lives in Markdown. You keep the files. Nothing is sent to a server.

---

Quick Start

Prerequisites

• Claude Code installed (npm install -g @anthropic-ai/claude-code)
• A Claude API key (Sonnet 4.5 or higher recommended)

Setup

1. Use this template

Click "Use this template" on GitHub, or clone it:

git clone https://github.com/pradnk/lore.git my-second-brain
cd my-second-brain

2. Open in Claude Code

claude

3. Edit your goals

Open brain/self/goals.md and replace the placeholder goals with your own. Be specific — the more concrete your goals and measurable signals, the better /learn performs.

4. Edit your profile (optional)

Open brain/self/profile.md to set your background and priority lens. This helps /review and /ask give more personalised answers.

5. Run your first learn session

/learn

Claude will read your goals, search for free resources, write wiki articles, and fetch a trivia fact.

---

Commands

Command	What It Does
/learn [topic]	Fetches free resources for your goals; writes wiki articles; fetches daily trivia
/ingest	Processes files in brain/raw/inbox/ into wiki articles and journal entries
/assess [topic]	Interactive knowledge quiz on your wiki articles; tracks weak areas
/ask <question>	Answers a question using only what your brain contains
/lint	Health-checks the wiki: orphans, gaps, stale index, duplicates
/reflect [notes]	Writes today's journal entry: decisions, what worked, open loops
/review [weekly|monthly|quarterly]	Progress review from journals + goals
/ngo <url>	Profiles an NGO or nonprofit from its website

---

Directory Layout

brain/
├── CLAUDE.md          ← operating manual (read this first)
├── wiki/              ← agent-written knowledge articles
│   ├── INDEX.md       ← one line per article; always current
│   └── ngo/           ← NGO profiles (excluded from /assess)
├── raw/               ← source material — APPEND-ONLY
│   └── inbox/         ← drop anything here: links, notes, PDFs
├── self/              ← your personal model (keep private)
│   ├── goals.md       ← what you're working toward
│   ├── profile.md     ← your values and working style
│   ├── journal/       ← daily episodic memory
│   └── ...
└── outputs/           ← query answers and reports

---

Capturing Content

Manually: Drop any .md file into brain/raw/inbox/ and run /ingest.

Via Discord (optional): Set up the Claude Code Discord MCP to DM yourself links and have them automatically pulled in during /ingest.

---

Privacy

• brain/self/ contains your personal journal and goals — do not commit this to a public repo unless you intend to.
• The .gitignore has brain/self/journal/, brain/self/goals.md, and brain/self/profile.md commented out. Uncomment those lines if you want to exclude them from git.
• brain/raw/ is append-only by design — Claude never edits source files.

---

Philosophy

• Goals-first — every /learn run starts from your goals, not a random topic feed.
• You own the files — Markdown, no proprietary format, no lock-in.
• Agent does the work — you capture intent; Claude does research, writing, and quizzing.
• Backlinked wiki — articles connect to each other with [[slug]] backlinks; /lint keeps them healthy.

---

License

MIT — do whatever you want with it.