Recallr

Recallr — turn photos and PDFs into compact study sheets and AI-generated quizzes to practise active recall.

---

Table of Contents

• What is Recallr?
• Screenshots
• Key Implemented Features
• Planned / Upcoming Features
• Architecture & Tech Stack
• Setup / Build
• Running the App / Quick Usage
• Contributing
• Changelog
• License

---

What is Recallr?

Recallr converts user-uploaded photos and PDFs into structured learn sheets using an OpenAI-powered pipeline and can generate interactive AI quizzes (chat-style) from that content so users can test and reinforce their knowledge.

Typical workflow:

• Upload photos / PDFs of notes or slides
• Generate a learn sheet (AI summarizes and formats content)
• Start an AI-powered chat/quiz to test knowledge based on the learn sheet

---

Screenshots

Learn Sheet	Learn Sheet View

Learn Sheet Chat	Settings View

Key Implemented Features

• 📚 Course & Learn Sheet Management – organize your subjects and generated study sheets
• 📄 File Import – import images (PNG, JPG, WEBP) and PDFs as source material
• 🤖 AI-Powered Summaries – automatically generate learn sheets from your uploaded files
• 💬 Interactive AI Chat & Quizzes – ask questions and get quizzed on your material with streaming responses
• 🕘 Chat History – previous conversations are saved per learn sheet
• 🎨 Light/Dark Theme – Fluent UI-based theming
• 🌐 Multi-Language Support – German & English
• ⚙️ Customizable AI Behavior – choose OpenAI model, summary style, difficulty, and question type
• 🔒 Encrypted API Key Storage – your OpenAI key is encrypted at rest via Microsoft.AspNetCore.DataProtection

Notes from code:

• Settings and courses are stored as plain JSON files in the user's ApplicationData folder (the OpenAI key itself is encrypted before being written; everything else is stored as-is).

---

Planned / Upcoming Features

Planned items (not yet implemented in code):

• Knowledge status view for learn sheets (tracking how well the user knows each topic)
• Adjustable font size for the learn sheet view (UI control to change font rendering)
• Interactive multiple-choice picker for quiz questions in the AI chat
• Error handling with custom error overlay

---

Architecture & Tech Stack

• UI: Avalonia (MVVM pattern)
• Language / Runtime: C# / .NET 10
• MVVM: CommunityToolkit.Mvvm
• AI: OpenAI .NET SDK (OpenAI.Chat integration)
• Markdown rendering: Markdown.Avalonia
• PDF handling: PDFsharp
• Charts: LiveChartsCore
• DI: Microsoft.Extensions.DependencyInjection (basic service registration)

Code layout highlights:

• App startup: Program.cs (creates %APPDATA%/Recallr/Config files if missing)
• Views: /Views (CourseView, LearningsheetView, LearningsheetDetailedView, SettingsView)
• ViewModels: /ViewModels (MainViewModel, CourseViewModel, LearningsheetDetailedViewModel, SettingsViewModel)
• Services: /Models/Services (AIService, SettingsService, CoursesService, ChatStorageService, FilePickerService, PdfService)
• Models: /Models/Configuration and /Models/Models

---

Setup / Build

Prerequisites:

• .NET 10 SDK (or newer)

Supported platforms: Windows and Linux are actively tested. macOS is not currently tested and may not work out of the box (untested, not officially supported yet).

Restore and run locally:

# from repository root
dotnet restore
dotnet build
dotnet run --project Recallr.csproj

Notes:

• The project references packages in the .csproj (Avalonia 12.1.0, OpenAI 2.12.0, Markdown.Avalonia, LiveChartsCore, PDFsharp, CommunityToolkit.Mvvm).
• No extra database required — the app stores JSON files under the OS ApplicationData folder (see FileSystemPaths in source).

---

Running the App / Quick Usage

1. Launch the app (dotnet run or your platform-specific build).
2. Open settings and paste your OpenAI API key into the key field, set profile name/email and preferred theme.
   • The key is encrypted before being saved to %APPDATA%/Recallr/Config/ClientSettings.json.
3. Create a course and open it.
4. Create a learn sheet and use the file picker to add files.
5. Click "Create study sheet" to generate the summary. The generated text is saved as learnsheet.md in the learn sheet folder.
6. Open the learn sheet chat to start AI-powered quizzes / active-recall practice. Chat is streamed and saved to per-learn-sheet JSON logs.

Paths used by the app (OS-dependent):

• Settings: %APPDATA%/Recallr/Config/ClientSettings.json
• Courses: %APPDATA%/Recallr/Config/ClientCourses.json
• Course files: %APPDATA%/Recallr/Courses/{courseId}/{learnsheetId}/
• Chat logs: %APPDATA%/Recallr/ChatLogs/{courseId}/{learnsheetId}.json

---

Contributing

This project is in early, active development (v0.2.0). Contributions, bug reports and feedback are welcome — open an issue or submit a pull request. Prefer small, focused PRs and describe breaking changes clearly.

Before contributing:

• Run dotnet restore and ensure the app builds locally.
• If changing settings storage or secrets handling, include migration notes.

---

Changelog

See CHANGELOG.md for the full version history.

---

License

This repository includes a LICENSE file: MIT License. See LICENSE for details.

---

If anything in this README is unclear or you'd like it adjusted (language, example screenshots, additional sections), say which parts to expand and a preferred tone (concise, tutorial, or marketing).