Turn AI chats into mastery.

A chatbot can explain a topic and draft ten questions in seconds. A chat transcript is still a poor place to learn from them. It does not keep an answer key, show which questions you got wrong or give you a reason to come back tomorrow.

QuizCaddy takes the multiple-choice questions your chatbot writes and gives them a stable home. You answer one at a time, check your choice against the stored answer, and read the worked solution when you are ready.

• Import a finished batch through MCP or paste the JSON into the Import page.
• Practise a deck, or a folder and its subfolders.
• Keep every checked answer and return to the questions that need another pass.

• Import a JSON deckThe complete schema, limits and a prompt that produces valid JSON.
• Connect through MCPSend questions from Claude, ChatGPT, Grok or another MCP client.

AI output gets messy when every topic lands in a new chat. QuizCaddy groups questions into decks with a topic and source, then adds optional subject, difficulty and exam labels.

On the paid plan, folder paths add a syllabus shape: Biology / Photosynthesis, Chemistry / Organic mechanisms or History / Cold War. Reuse the same path for the next batch and the new deck joins the existing chapter.

• Use the topic for the deck name.
• Use subject and exam labels when you want to filter practice later.
• Use folder paths for the syllabus, not for exam names.

Open a practice session and answer before you reveal the solution. QuizCaddy checks the choice immediately and records the attempt. The worked solution is attached to the question, so you can inspect the reasoning while the question is still fresh.

Finish the deck, then start another session when you want another pass. Repetition is part of the design, not a workaround.

• Deck practice for one import or one topic.
• Folder practice when you want a whole chapter, including subfolders.
• Mistake revisions when you want the questions that did not stick.
• Timed exams when you need to practise pacing as well as recall.

A wrong answer is more useful than a clean score. Every checked answer is recorded, and incorrect answers enter your mistake history. The paid mistake bank turns that history into a revision queue with due dates and manual mastery.

You can revise up to 50 due mistakes in a session, mark a question mastered, remove it, or restore it later. The goal is not to keep every mistake forever; it is to notice when the reasoning has changed.

Insights adds the wider pattern: accuracy, active days, streaks, topic volume and weakest topics. It counts checked answers, including repeated questions in different sessions, and exam answers appear after the exam is finalised.

When the exam is the target, use Exam mode. Choose a deck, folder, mistakes or topic; set 5 to 200 questions and a 5 to 360 minute limit; choose no negative marking, +4 / −1, or a custom positive and penalty scheme.

The runner keeps a server-anchored countdown, lets you flag questions and saves draft choices. After submission you get score, accuracy of attempted questions, time used, a topic breakdown and a review of every answer.

This is a study tool, not a proctored exam. Leaving the page does not stop the timer, and answers that do not reach the server before the deadline cannot count.

You do not need a tutor to make the loop work. You need a small, repeatable routine and questions that stay in one place.

1. 1

   Pick the weak spot

   Ask your chatbot for questions on one topic, or use Weak areas to choose a topic from your own history.

2. 2

   Import the batch

   Use the MCP connector or paste the JSON. Check the deck title, topic and folder path before you save.

3. 3

   Answer before you peek

   Work through the deck without notes. Read the solution after every check, not only the wrong ones.

4. 4

   Return to the mistakes

   Revise due mistakes, mark what has clicked and remove what no longer needs attention.

5. 5

   Take a timed run

   Use Exam mode near the end of the week to practise pacing and marking.

6. 6

   Let the data choose next week

   Open Insights, find the weakest topics and repeat the loop.