A private study deck that knows which lecture and page you have reached.
Anki turns reached course material and book pages into cited study cards. It reads source passages from Cortex and keeps cards, reading checkpoints, and FSRS reviews in local SQLite. The phone PWA runs on your Mac and opens through a private Tailscale address.
Thirty distinct reviews per day, five recall grades, swipe controls, and undo keep daily study bounded. ChatGPT can read eligible material and submit cited cards through 14 MCP tools. This is Pablo's own study app, separate from the Anki project at apps.ankiweb.net.
Requires Node.js, macOS, Tailscale, and local Cortex for course and book context.
npm install
npm run build
npm test
python3 ops/install_local.pyOpen http://127.0.0.1:3464. The installer starts Anki at login and serves tailnet HTTPS on port 8444. Data stays in ~/Library/Application Support/Anki across updates. The app has no public study database endpoint.
npm run mcp starts the stdio server. For ChatGPT, use a private Secure MCP Tunnel and follow the setup guide. A local MCP connection does not establish native ChatGPT phone support.
The approved symbol, wordmark, lockup, and app icon are generated from the editable curves in ops/generate_brand_icons.mjs. Run npm run brand:icons with librsvg installed to regenerate every SVG, the README logo, favicon, and install icons.
