Cleans up your pasted MCQ text to the required format (Qn., options with “-”, and an ANSWER KEY). If missing, it assigns a single correct answer per question. Output replaces the text above.
Options
No file chosen
Types
No key set
AI Output Preview
Upload a .docx, .pdf, or .pptx. Gemini will conservatively extract content and return a set of
questions formatted for this tool (either MCQ text with an ANSWER KEY, or a JSON bank with { items: [...] }).
Saved Sets
0
Total: 0.0 KB
Flashcards & SRS
No deck loaded.
00:00
Question
Answer
Session: 0 / 0
Preview
Parser Check
No lint results yet.
Format Information
Standard MCQ: Q1. Question text...
Options start with - (e.g., - Option A).
Add an ANSWER KEY at the end: 1. B 2. C
True/False:
Add [TF] to the question text.
Key: 2. True or 2. False.
Fill-in-the-Blank:
Add [FIB] to the question text.
Key: 3. Answer (multiple allowed: 3. answer | alt answer).
Matching:
Add [MATCH] to the question text.
Use lines like - Left Item -> Right Item.
Optional extras: - Extras: distractor1 | distractor2.
Advanced: JSON question banks are also supported.
Quick Start — Full Guide
This app is a single-file, fully local quiz engine. Your pasted document is parsed into a deck,
then you can quiz, export, save, and back up your data.
1) The core workflow (fast path)
Paste MCQs (or load a .txt file).
Click Parse & Preview.
Scan the Preview count and the Parser Check panel (fix issues if needed).
Click Start Quiz.
2) Input formats (what the parser expects)
MCQ:Q1. then options as - lines, then an ANSWER KEY at the end.
True/False: add [TF] to the question; key with True/False.
Fill blank: add [FIB] and use ____ in the prompt; key with one or more answers.
Matching: add [MATCH] and use Left -> Right pairs.
Advanced JSON: you can paste a JSON bank (array or { items: [...] }).
3) Parser Check (lint) — what it means
Errors usually mean the deck may behave incorrectly (e.g., invalid correct answer index, missing options).
Warnings are common for messy numbering or missing blanks.
Renumber makes question numbers 1..N for the current parsed deck.
Remove dupes drops repeated question numbers (keeps the first occurrence).
4) Quiz settings (Options)
Shuffle questions: changes order each run.
Shuffle answers: shuffles choices inside each question.
Instant feedback: shows correct/wrong immediately after selection.
5) Saving & reusing sets (Saved Sets)
Use Save Set to store the current pasted document for reuse later.
Saved items are stored locally in your browser (no server required).
6) Exports
Export PDF: prints the Preview deck (optional highlight correct answers).
Export Anki TSV: downloads a TSV with Front and Back columns (Anki import-ready).
Flashcards & SRS: a separate practice mode for spaced repetition (use Export CSV / Export Anki TSV as needed).
7) AI features (Gemini)
AI features use the global Gemini API key at the top of the page.
Reformat & Assign Answers (AI): fixes formatting and builds an answer key when missing.
Generate Like-Questions (AI): rewrites your deck into similar, format-identical items.
Validate Answers (AI): checks your answer key vs options and reports mismatches.
Tip: if your deck won’t parse cleanly, run the AI reformat first, then Parse & Preview.
8) Backup / Restore / Share (local device state)
Backup exports your browser state to a JSON file (Saved sets, saved textbooks, settings).
Restore can Merge (keep existing) or Overwrite (replace local state).
Include API keys is OFF by default for safety (recommended).
Share current doc exports only the current pasted document + quiz options (no keys, no libraries).
9) Troubleshooting checklist
If the quiz is empty: verify Preview shows items parsed.
If answers look wrong: verify your ANSWER KEY numbering matches the questions.
If AI buttons are disabled: enable the global key toggle and ensure the key is valid.
If a device/browser change is coming: export a backup first.
Want the shortest possible path? Use: Paste → Parse & Preview → Fix Parser Check (if needed) → Start Quiz.
Building questions…
Today's Random Quiz is about choosing...!
Preparing…
This dialog displays progress while the daily quiz is generated. Messages are also logged to the Generate MCQ debug output.
All Questions
Summary of your quiz performance. Use the analyzer below for AI feedback on missed questions and trends.
No key set
Saved Quiz Results
0
Total: 0.0 KB
Last Attempted Quiz
None yet
Content Library
Choose files to extract chapters and text into your Content Library. Scan to preview, then use Save to store items; use the Saved Content controls to show, delete, or clear entries.
Depth
-
Pages
-
File: (none)
Pages: 0
Estimated: -
Activity / Progress
Content Preview
(no TOC scanned)
Saved Content
(none)▾
Total: 0.0 KB
(select saved content and click Show)
No key set
Validates each ANSWER KEY entry against its question options. Outputs a single-line summary (CORRECT / INCORRECT counts) followed by only the incorrectly keyed questions and their correct answers into the box below.
No key set
Carefully replaces your pasted set with new, format-identical items. Uses conservative
rewriting; keeps same count, structure, and single correct answer per question.
Utilities
· Backup your local data, share docs, and run diagnostics.
Backup & Restore
Exports your local browser state (Saved sets, saved textbooks, settings). Use this before switching devices/browsers.
Included in backup
Restore mode
No file chosen
About AI Elements (Gemini 2.5 Flash)
AI requests are sent directly from your browser to Google Gemini when enabled. Your key is stored locally on this device only if you turn on “remember”, and you can exclude keys from backups.
The rationale shown in the panel is generated by Google’s Gemini model. While useful for learning,
it may occasionally misidentify the correct option or provide incomplete justification. Reasons include:
Non-determinism & sampling: temperature/top-p sampling can yield different
rationales across runs, including confident but incorrect claims.
Prompt/format drift: if the question/options aren’t parsed exactly as intended,
the model may reason over the wrong input.
Hallucinations & shortcut bias: large models sometimes invent facts or over-rely
on surface cues rather than domain-accurate reasoning.
Answer-key disagreement: if your key or parsing is off, the model could be “right”
while disagreeing with the displayed key (or vice-versa).
For guidance on safety, evaluation, and reliability, see:
API Docs