Lore uses Claude to research your goals, write knowledge articles, and quiz you on what you've learned — all in plain Markdown files you own.
Not a note-taking app. Not a reader. An agent that builds knowledge for you — organised by what you're trying to achieve.
Every /learn run starts from your goals.md — not a random topic feed. Claude reads what you're working toward and finds resources aligned to it.
You set intent. Claude searches, reads, summarises, writes wiki articles, and links them together. You read, not research.
Everything is plain Markdown in a git repo. No proprietary format, no lock-in, no server storing your thoughts. Fork it, back it up, share it.
Articles connect to each other with [[slug]] backlinks. /lint keeps the graph healthy — finds orphans, stale index lines, and missing links.
You need Claude Code and an Anthropic API key. Everything else is already in the template.
Claude Code is the CLI that powers Lore's agent commands.
npm install -g @anthropic-ai/claude-code
Click "Use this template" on GitHub, or clone directly.
git clone https://github.com/pradnk/lore my-brain cd my-brain
Edit brain/self/goals.md. Add 1–3 specific goals with a Why and measurable signals. The more concrete, the better /learn performs.
# Goals ## Goal 1 — Learn machine learning - Why: Career shift into ML engineering - Signals: Complete fast.ai; build 2 projects - Status: in progress
Edit brain/self/profile.md — your background, working style, and values. This gives /ask and /review better context about who you are and how you think.
# Profile ## Background Software engineer, 5 years in fintech. New to ML; strong in Python and system design. ## Working Style Prefer depth over breadth. Learn by building. ## Values Continuous improvement, teaching others.
Launch Claude Code in your brain directory. Run /learn and watch it go.
claude > /learn
Each command is a Claude Code slash command stored in .claude/commands/. You can read, edit, and extend them.
Fetches free resources aligned to your goals, writes wiki articles, and adds a daily trivia fact.
Processes files in brain/raw/inbox/ — articles, PDFs, Discord DMs — into wiki articles and journal entries.
Interactive knowledge quiz on your wiki. Tracks weak areas and scores per article over time.
Answers any question using only what your brain contains. Cites sources and flags gaps.
Health-checks the wiki: finds orphans, stale index lines, missing backlinks, and contradictions.
Writes today's journal entry: decisions made, what worked, open loops, energy signals.
Progress review against your goals. Weekly consolidates journal; monthly proposes profile updates.
Profiles an NGO or nonprofit from its website into a structured wiki article.
Everything lives in one brain/ directory. Plain Markdown. In git. Yours.
brain/ ├── CLAUDE.md ← operating manual ├── wiki/ │ ├── INDEX.md ← one line per article │ ├── mutual-funds.md │ ├── jlpt-n5.md │ └── ngo/ ← NGO profiles ├── raw/ ← append-only source material │ └── inbox/ ← drop anything here ├── self/ ← your personal model │ ├── goals.md ← what you're working toward │ ├── profile.md ← values + priority lens │ └── journal/ ← daily episodic memory └── outputs/ ← answers and reports
Clone the template, set your goals, run /learn.
MIT license · No account required · Your data stays local