Learning Loop

Inspiration

Practice is more actionable when the next question and prior attempts are visible. Learning Loop explores a compact, user-authored practice cycle.

What it does

It stores questions with difficulty, selects the least-mastered question for a topic (breaking ties toward higher difficulty), scores a case-insensitive exact answer, and reports persistent attempts, accuracy, and average mastery.

How we built it

The implementation is a local Python MCP server tested via Streamable HTTP (MCP 2025-11-25), with SQLite-backed state and typed tools. It uses deterministic matching rather than semantic grading. A static browser replay, when used, is a JavaScript/localStorage equivalent and is not connected to MCP or Alexa.

Challenges

The selection rule needs a clear definition of mastery, while answer scoring must not imply understanding beyond an exact normalized match.

What we learned

Simple persistence can make progress inspectable without claiming adaptive tutoring or generated content.

Accomplishments

Local tests and a fresh Streamable HTTP transcript verify question creation, selection, answer scoring, and progress reporting.

What's next

Add richer grading reviewed by educators, spaced scheduling, export and privacy controls, and optional host integration without overstating learning efficacy.

This is a local single-user prototype: no cloud or LLM claim, no generated curriculum, semantic grading, Alexa/Echo integration, or efficacy study.

Built With

• css3
• fastmcp
• html5
• javascript
• json-schema
• localstorage
• model-context-protocol
• pytest
• python
• sqlite
• streamable-http