Type & Comprehend — AI Curriculum

Part of datainsight.at · Live demo

A typed learning tool for data and AI engineers. Read a topic essay, then type your answers to comprehension questions — reinforcing understanding through active recall and deliberate typing.

---

How It Works

1. Pick a category (e.g. AI Fundamentals, Data Engineering, Agent Engineering)
2. Select a topic — each topic has a full essay in DE and EN
3. Read the essay, then answer 3 comprehension questions by typing
4. Your answers are checked by an LLM for understanding (requires API key)
5. Progress is tracked per topic across sessions

The app tracks WPM, accuracy, time, and completion percentage as you type.

---

Content

All content lives in content.json — no other files need editing to add topics.

Structure

{
  "categories": [
    {
      "id": "category-slug",
      "de": { "title": "...", "icon": "🤖" },
      "en": { "title": "...", "icon": "🤖" },
      "topics": [
        {
          "id": "topic-slug",
          "de": { "title": "...", "body": "...", "questions": ["..."] },
          "en": { "title": "...", "body": "...", "questions": ["..."] }
        }
      ]
    }
  ]
}

Adding a Topic

1. Open content.json
2. Find the target category (or add a new one)
3. Add a new object to topics[] with id, de, and en fields
4. Each topic needs: title, body (the essay), and questions (array of 3 strings)

No code changes needed — app.js renders everything from content.json.

---

Current Categories

Category	Topics
🤖 AI Fundamentals	LLMs, Transformers, Embeddings, RAG, Agents
🗄️ Data Engineering	Pipelines, Lakehouse, dbt, Orchestration
🔗 Agent Engineering	MCP, HITL, AHI, ADPL

---

Files

File	Responsibility
index.html	App shell — 4 screens: home, category, typing, results
app.js	Screen routing, typing engine, LLM answer checking, progress persistence
style.css	All component styles (self-contained, no shared-theme dependency)
content.json	All topics, essays, and questions in DE + EN

---

Running Locally

No build step. Open index.html directly in a browser, or serve from the prompt_engineer root:

cd /path/to/prompt_engineer
python3 -m http.server 8080
# open http://localhost:8080/tools/type_comprehend/

To enable LLM answer checking, configure an API key in the app settings (OpenAI-compatible endpoint).

---

Part of the DE Prompt Toolkit

This tool is embedded in datainsight.at as a submodule of datainsightat/prompt_engineer.

---

License

MIT