educationai

Features

• •FSRS-6 checked against the reference: the tests replay 2,400 reviews computed by py-fsrs 6.3.2 over eight configurations and compare stability, difficulty, due date and retrievability after every review
• •Learning and relearning steps, desired retention, maximum interval, interval fuzz and custom weights (21 from FSRS-6 or 19 from FSRS-5); a preview of all four ratings for the answer buttons
• •SM-2 as published, and a conversion of SM-2 cards to FSRS that keeps each card's due date
• •Typed-answer grading that ignores case, accents, width and punctuation without depending on the device language, forgives typos by answer length and does bounded work on hostile input
• •Multiple-choice and typed quizzes built from cards; each result suggests a rating for the scheduler, and a saved quiz is graded again when it is loaded
• •A study queue with daily limits, a per-day forecast, retention statistics, streaks counted in the learner's calendar days and a difficulty adjuster that keeps accuracy in a target band
• •Sendable, Codable value types: every function takes the date it works with, randomness comes from a generator you can seed, and stored values are checked before use instead of crashing
• •No dependencies, only Foundation; DocC reference with four guides and a StudyDeck example app for iOS and macOS

Quickstart

1struct FlashcardsView: View {
2    @State private var cards = [
3        Flashcard(front: "Capital of Japan", back: "Tokyo"),
4        Flashcard(front: "Capital of Kenya", back: "Nairobi"),
5        Flashcard(front: "Capital of Peru", back: "Lima")
6    ]
7    @State private var showsAnswer = false
8    private let scheduler = FSRSScheduler.default
9 
10    var body: some View {
11        if let card = StudyQueue.cards(from: cards).first,
12           let index = cards.firstIndex(where: { $0.id == card.id }) {
13            VStack(spacing: 24) {
14                Text(card.front).font(.title)
15                if showsAnswer {
16                    Text(card.back).font(.title2).bold()
17                    HStack {
18                        ForEach(Rating.allCases, id: \.self) { rating in
19                            Button(String(describing: rating).capitalized) {
20                                if let outcome = try? scheduler.review(card.review, rating: rating) {
21                                    cards[index].review = outcome.state
22                                }
23                                showsAnswer = false
24                            }
25                            .buttonStyle(.bordered)
26                        }
27                    }
28                } else {
29                    Button("Show Answer") { showsAnswer = true }
30                        .buttonStyle(.borderedProminent)
31                }
32            }
33            .padding()
34        } else {
35            Text("All done for now")
36        }
37    }
38}

Installation

1https://github.com/muhittincamdali/educationai.git

1dependencies: [
2    .package(url: "https://github.com/muhittincamdali/educationai.git", from: "2.0.0")
3],
4targets: [
5    .target(name: "YourApp", dependencies: [
6        .product(name: "EducationAI", package: "educationai")
7    ])
8]

Screenshots

Timeline

1. August 2025

   Repository opened; 1.0.0 tagged on 7 August: an EducationAI class with learning models, depending on Alamofire, RxSwift, SnapKit and SDWebImage

2. February 2026

   Rewritten without dependencies as SM-2 scheduler, quiz, progress, recommendation and gamification engines

3. June 2026

   A dependency on a local folder (../SwiftAI) and placeholder "neural tutor" and "enterprise" types were added; the package no longer resolved for anyone else

4. October 15, 2026

   2.0.0: rebuilt around an FSRS-6 scheduler that matches py-fsrs 6.3.2; the June code and every 1.x type removed

Links

Related Article