Known concepts live in `data/data-engineering-knowledge.json`, which means you can read it, edit it and check it into a repo — and also that there is one profile per checkout and no concurrency story. If Cursor does not pick up the server's prompt automatically, the README's fallback is to paste the same guidance into a project rule file and reference it in your request.
A tutor server that fetches recent data-engineering developments and filters them against a stored profile of what you already know, so the assistant tells you about the parts that are new to you.
- `de_tutor_get_updates` queries Perplexity through OpenRouter for recent data-engineering news, patterns and technologies
- `de_tutor_read_memory` returns the concepts already marked as known
- `de_tutor_write_memory` marks a concept as known or unknown, taking a `concept` and a `known` boolean
- A `data_engineering_knowledge_memory` resource backing all of it — a JSON object mapping concept names to booleans
- A `data-engineering-tutor-guidance` prompt that tells the client the intended order: fetch updates, read memory, present what is new, then update memory from your answers
Build from source: `npm install` then `npm run build`, and launch `build/index.js` with node using an absolute path. The project is data-engineering-tutor, version 1.0.0, and the binary name is `de-tutor`. Node.js v18 or later. `de_tutor_get_updates` needs an OpenRouter API key in `OPENROUTER_API_KEY`, either in a `.env` file at the project root or in the client's env block. The other two tools need no key.