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
- You set goals in
brain/self/goals.md - Run
/learn→ Claude finds free resources aligned to your goals and writes wiki articles - Run
/ask what should I focus on this week?→ Claude answers from your wiki - Run
/assess→ interactive knowledge quiz on what you've learned - Run
/reflect→ daily journal entry consolidating decisions and signals
Everything lives in Markdown. You keep the files. Nothing is sent to a server.
- Claude Code installed (
npm install -g @anthropic-ai/claude-code) - A Claude API key (Sonnet 4.5 or higher recommended)
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-brain2. Open in Claude Code
claude3. 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.
| 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 |
brain/
├── CLAUDE.md ← operating manual (read this first)
├── wiki/ ← agent-written knowledge articles
│ └── INDEX.md ← one line per article; always current
├── 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
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.
brain/self/contains your personal journal and goals — do not commit this to a public repo unless you intend to.- The
.gitignorehasbrain/self/journal/,brain/self/goals.md, andbrain/self/profile.mdcommented out. Uncomment those lines if you want to exclude them from git. brain/raw/is append-only by design — Claude never edits source files.
- Goals-first — every
/learnrun 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;/lintkeeps them healthy.
MIT — do whatever you want with it.