SYSTEM CONTEXT FOR AI (STRICT PRESERVATION) CRITICAL INSTRUCTION: This document is the single, authoritative Source of Truth for "IYmtg," an iOS application designed to identify, grade, and insure Magic: The Gathering cards.

Version Authority: This Version 1.0.0 supersedes all previous iterations.
Architecture Mandate: Any future updates must strictly adhere to the defined pipeline: Vector Fingerprinting (Identity) -> OCR (Validation) -> ML Analysis (Condition/Foil).
Preservation Protocol: Do not summarize, truncate, or remove sections of this manual during review.

Development Roadmap

This is the complete sequence of steps to go from source code to a working app. Complete them in order.

Step	Task	Platform	Status
1	Workspace setup & visual assets	Any	—
2	Build `IYmtg_Builder_Mac`	Mac only	⚠️ Not written yet
3	Generate `cards.json` database	Mac only	⚠️ Depends on Step 2
4	Create Xcode project from source	Mac only	—
5	Collect ML training images	Any (physical cards)	—
6	Train ML models in Create ML	Mac only	—
7	Configure Firebase (optional)	Any	—
8	Final configuration & testing	Mac only	—
9	App Store submission	Mac only	—

What You Can Do Without a Mac

Edit source code
Run Python automation scripts (fetch_set_symbols.py, generate_images.py)
Collect and sort ML training images into IYmtg_Training/
Acquire physical cards from the shopping lists

What Requires a Mac

Everything else. Apple's Vision framework (used to generate card fingerprints) and Create ML (used to train models) are macOS-only. The Xcode project also lives on your Mac.

Part 1: App Store Listing

1. Metadata

App Name: IYmtg: Card Scanner & Insurance
Subtitle: Identify, Grade & Insure Magic
Category: Reference / Utilities
Keywords: magic,gathering,scanner,tcg,card,price,insurance,manager,grade,foil,mtg,free,offline
Device Orientation: Strictly lock to Portrait in Xcode.

2. Description

Headline: The Easiest Way to Insure Your Magic Collection.

Body: Your Magic: The Gathering collection represents years of history and passion. Losing it to theft, fire, or disaster is a nightmare scenario. IYmtg is the first app built specifically to make insuring your collection simple, fast, and accurate.

Forget complex spreadsheets and manual entry. Just point your camera, and IYmtg handles the rest. It identifies the card, grades the condition, detects foiling, and fetches the market price instantly. When you're done, one tap generates a professional Insurance Schedule PDF ready for your agent.

Why IYmtg?

📄 Insurance Ready: Generate a timestamped, itemized PDF Schedule in seconds.
⚡ Effortless Scanning: Auto-detects Set, Condition, and Foil type (including Etched, Galaxy, and more).
🔒 Private & Secure: Your data is backed up, but your images stay private in iCloud.
✅ Simple & Clean: No ads, no subscriptions, just a powerful tool for collectors.

Development Transparency: This application's code and visual assets were developed with the assistance of Artificial Intelligence. This modern approach allows us to deliver a sophisticated, high-performance tool dedicated to a single goal: helping collectors manage, grade, and insure their valuable history with precision and ease.

Community Data Initiative: Help us make IYmtg smarter! If you find a card that scans incorrectly, you can correct it in the app. When you do, you'll have the option to securely send that image to our training database. Your contributions directly improve the AI models for the entire community.

Features:

Insurance Reports: Export your entire collection to a PDF ready for your insurance agent.
Collection Valuation: Monitor the total value of your collection with real-time market data.
Smart Scanning: Identify cards, foils, and condition automatically.
Cloud Sync: Keep your collection safe and accessible across your devices.
Offline Access: Scan and manage your cards even without an internet connection.
Market Data: Switch between major pricing sources (TCGPlayer & Cardmarket).
Export Options: Also supports CSV and digital deck formats for other uses.

Part 2: Workspace & Assets

Step 1: Workspace Setup

Create the master folder in your preferred location (Desktop, OneDrive, or any synced drive).
Ensure the folder is synced with a cloud backup service (OneDrive, Google Drive, iCloud Drive, etc.).
Organize your sub-folders exactly as shown below:

IYmtg_Master/
├── IYmtg_App_iOS/       (The iOS App Source Code)
├── IYmtg_Builder_Mac/   (The Card Database Builder — Mac app)
├── IYmtg_Training/      (ML Image Data)
└── IYmtg_Automation/    (Python/Shell Scripts)

Step 2: Visual Assets

Place the following assets in Assets.xcassets in the Xcode project.

Important: AI tools often generate large files (e.g., 2048x2048). You must resize and crop the results to the dimensions listed below. For the AppIcon, exact 1024x1024 dimensions are mandatory.

Asset Name	Dimensions	Description	Gemini Generation Prompt
AppIcon	1024x1024	App Icon.	"A high-quality iOS app icon. A stylized neon green cybernetic eye scanning a dark, mystical trading card silhouette. Dark purple and black background. Minimalist, sleek, modern technology meets fantasy magic. No text. Square aspect ratio."
logo_header	300x80	Header Logo.	"A typographic logo for an app named 'IYmtg'. Horizontal layout. Neon green text, futuristic sans-serif font. Dark background. The text should be glowing. High contrast. Aspect ratio 4:1."
scanner_frame	600x800	Viewfinder.	"A HUD viewfinder overlay for a camera app. Glowing white bracket corners. Thin, high-tech lines connecting corners. Center is empty. Sci-fi interface style. Pure white lines on a solid black background. Aspect ratio 3:4."
empty_library	800x800	Empty State.	"Isometric 3D render of a clean, empty wooden desk. A single Magic: The Gathering style card sits in the center. Soft warm lighting. Minimalist design. High resolution. No text. Square aspect ratio."
share_watermark	400x100	Watermark.	"A watermark logo text 'Verified by IYmtg'. White text with a checkmark icon. Clean, bold font. Solid black background. Professional verification seal style. Aspect ratio 4:1."
card_placeholder	600x840	Loading State.	"A generic trading card back design. Grey and silver swirl pattern. Mystical and abstract. No text. Aspect ratio 2.5:3.5."

Automated Generation (Recommended)

Setup:

Get a Gemini API Key: You will need an API key from Google AI Studio.
Set the API Key: Open IYmtg_Automation/generate_images.py and set your API key in the configuration section.
Install dependencies:
```
pip install requests pillow
```

Usage:

python3 IYmtg_Automation/generate_images.py

The generated images will be saved in Raw_Assets and resized images in Ready_Assets.

Manual Resizing (If You Already Have Images)

Setup: Ensure Python is installed and run pip install Pillow.

Generate Placeholders (Optional):

python3 IYmtg_Automation/generate_placeholders.py

Place Images: Save your real AI results into Raw_Assets, named exactly as listed above (e.g., AppIcon.png).

Run:

python3 IYmtg_Automation/resize_assets.py

Result: Xcode-ready images will be in Ready_Assets. Drag them into Assets.xcassets.

Part 3: Card Database (`cards.json`) — Mac Required

This is the most critical file in the project. The app cannot identify any cards without it. It is a JSON file bundled inside the app containing a fingerprint (a mathematical representation) of every Magic card, generated from card images using Apple's Vision framework.

What `cards.json` Contains

Each entry in the file represents one unique card printing and contains:

Card name, set code, and collector number
Whether the card has a foil or serialized printing
Pricing data
A VNFeaturePrintObservation (binary blob) — the visual fingerprint used for identification

Step 1: Write `IYmtg_Builder_Mac`

IYmtg_Builder_Mac/ is currently empty. This Mac command-line tool needs to be built before cards.json can be generated. It must:

Fetch the complete card list from the Scryfall API (https://api.scryfall.com/bulk-data → default_cards dataset)
Download a card image for each unique printing
Run VNGenerateImageFeaturePrintRequest (Apple Vision) on each image to produce a fingerprint
Archive the fingerprint using NSKeyedArchiver into a Data blob
Write all entries to cards.json using the CardFingerprint model defined in IYmtg_App_iOS/Data/Models/Card.swift
Place the output at IYmtg_App_iOS/cards.json

Data model reference (CardFingerprint in IYmtg_App_iOS/Data/Models/Card.swift):

struct CardFingerprint: Codable {
    let id: UUID
    let name: String
    let setCode: String
    let collectorNumber: String
    let hasFoilPrinting: Bool
    let hasSerializedPrinting: Bool?
    let priceScanned: Double?
    let featureData: Data   // NSKeyedArchiver-encoded VNFeaturePrintObservation
}

To build IYmtg_Builder_Mac: Give this README section to an AI (Claude or Gemini) along with the CardFingerprint struct and ask it to write a Swift command-line tool for macOS. The tool is straightforward — it is a single-purpose script that runs once and can take several hours to complete due to the volume of Scryfall image downloads.

Step 2: Run the Builder

On your Mac, build and run IYmtg_Builder_Mac. The weekly_update.sh script automates this:

chmod +x IYmtg_Automation/weekly_update.sh
./IYmtg_Automation/weekly_update.sh

This script:

Builds the builder app using xcodebuild
Runs it (this takes time — Scryfall has ~100,000+ card printings)
Moves the output cards.json into IYmtg_App_iOS/ ready to be bundled

Run this script periodically (e.g., weekly) to pick up newly released sets.

Step 3: Add `cards.json` to Xcode

After the builder runs, cards.json will be at IYmtg_App_iOS/cards.json.

In Xcode:

Drag cards.json into the project navigator under IYmtg_App_iOS/
Ensure "Add to target: IYmtg" is checked so it is bundled inside the app

Part 4: Xcode Project Setup — Mac Required

The source code in IYmtg_App_iOS/ is complete. Follow these steps to create the Xcode project on your Mac.

Step 1: Create the Project

Open Xcode → File → New → Project
Choose iOS → App
Set:
- Product Name: IYmtg
- Bundle Identifier: com.<yourname>.iymtg (you choose this — note it down, you'll need it for iCloud)
- Interface: SwiftUI
- Language: Swift
- Storage: None (we use SwiftData manually)
Save the project inside IYmtg_App_iOS/ — this places the .xcodeproj alongside the source files.

Step 2: Add Source Files

In the Xcode project navigator, right-click the IYmtg group → Add Files to "IYmtg"
Select all folders inside IYmtg_App_iOS/:
- Application/
- Data/
- Features/
- Services/
- Firebase/
- AppConfig.swift
- ContentView.swift
- IYmtgApp.swift
- IYmtgTests.swift
- cards.json (once generated)
Ensure "Copy items if needed" is unchecked (files are already in the right place) and "Create groups" is selected.

Step 3: Add Dependencies (Swift Package Manager)

File → Add Package Dependencies
Add the Firebase iOS SDK:
- URL: https://github.com/firebase/firebase-ios-sdk
- Add these libraries to your target: FirebaseCore, FirebaseFirestore, FirebaseAuth, FirebaseStorage

Step 4: Configure Signing & Capabilities

Select the project in the navigator → Signing & Capabilities tab
Set your Team and ensure Automatically manage signing is on
Set Minimum Deployments to iOS 17.0
Click + Capability and add:
- iCloud → enable CloudKit → add container iCloud.<your-bundle-id>
- Background Modes → enable Remote notifications
Lock orientation: General tab → Deployment Info → uncheck Landscape Left and Landscape Right

Step 5: Add Privacy Descriptions

In Info.plist, add these keys (Xcode will prompt on first run, but adding them manually avoids rejection):

Key	Value
`NSCameraUsageDescription`	`IYmtg uses the camera to scan and identify Magic: The Gathering cards.`
`NSPhotoLibraryUsageDescription`	`IYmtg can save card images to your photo library.`

Also add PrivacyInfo.xcprivacy to the app target to satisfy Apple's privacy manifest requirements (required for App Store submission as of 2024).

Step 6: Build and Run

Select a connected physical iPhone as the build target (camera features do not work in the simulator)
Press Cmd+R to build and run
On first launch the app will show "Database Missing" until cards.json is bundled (see Part 3)

Part 5: Machine Learning Training — Mac Required for Final Step

In-App Training Guide (v1.1.0+): The Library tab now includes a "?" button that opens a color-coded Training Guide. It shows every ML category with recommended image counts for three accuracy levels (Functional / Solid / High-Accuracy), so you can track collection progress directly from the app.

You do not need the app, Xcode, or a Mac to collect training images. All you need is physical cards and a phone camera. The only Mac-required step is the final model training in Create ML (Step 6).

The app ships and works without any trained models. Foil detection defaults to "None" and condition defaults to "NM". You can release a working app first and add models later via OTA update. Do not let missing training data block your first build.

Step 0: How to Create Training Images (No App Required)

This is the complete workflow for preparing training data on any computer.

What You Need

Physical Magic cards (see shopping lists in Steps 1–3)
A phone with a decent camera (iPhone, Android — anything works)
A plain background: white card stock or black felt works best
A free photo cropping tool:
- Windows: Photos app (built-in crop) or Paint
- Any platform: Squoosh (browser-based, free, no install)
- Bulk cropping: IrfanView (Windows, free) or XnConvert (cross-platform, free)

Photography Setup

For Foil Cards:

Place the card on a black background (black felt or black paper).
Use a single directional light source — a desk lamp or window at 45°.
Take 5 photos of the same card rotating it slightly between each shot so the light catches the foil at different angles. The AI must learn how the foil moves, not just how it looks flat.
Example angles: flat-on, 15° left tilt, 15° right tilt, 15° top tilt, 15° bottom tilt.

For Non-Foil Cards:

Place on white or grey background.
Even, diffused lighting (avoid strong reflections on the surface).
1–3 photos per card is sufficient.

For Condition/Damage:

Use raking light (light source almost parallel to the card surface) — this casts shadows that highlight scratches, dents, and bends far more clearly than direct light.
For edge whitening: photograph against a black background.
For chipping: photograph against a white background.
Take a close-up — fill the frame with the card.

Cropping Rule — Critical

The app scans cropped card images only (no table, no background, no hand visible). Your training images must match this exactly or the model will learn the wrong thing.

After photographing:

Open the photo in your cropping tool.
Crop tightly to the card border — include the full card frame but nothing outside it.
It does not need to be pixel-perfect. Within 5–10px of the edge is fine.
Save as .jpg at any reasonable resolution (at least 400×560px).

Naming and Sorting

File names do not matter — only the folder they are in matters. Save cropped images directly into the appropriate IYmtg_Training/ subfolder:

IYmtg_Training/Foil_Data/Etched/   ← drop etched foil photos here
IYmtg_Training/Foil_Data/Traditional/  ← drop traditional foil photos here
IYmtg_Training/Condition_Data/Edges/Whitening/  ← drop edge whitening photos here

How Many Images Do You Need?

Goal	Minimum	Recommended
Test that training works	10 per class	—
Functional model, limited accuracy	20 per class	—
Solid production model	30–50 per class	50+ per class
High-accuracy model	—	100+ per class

More is always better. Variety matters more than quantity — different cards, different lighting, different tilt angles.

Using Scryfall Images as a Supplement

For NonFoil training data you can download card images directly from Scryfall instead of photographing them. This is automated — run:

pip install requests pillow
python3 IYmtg_Automation/fetch_set_symbols.py

This downloads and crops set symbol images automatically. For general NonFoil card images, you can query the Scryfall API directly (https://api.scryfall.com/cards/random) and download the normal image URI. Downloaded Scryfall images are already cropped to the card frame and work well as NonFoil training data. Do not use Scryfall images for foil or damage training — they are flat renders with no foil or physical damage.

General Data Collection Protocol (Critical)

The app sends cropped images (just the card, no background) to the AI. Your training data must match this.

Capture: Take photos of the card on a contrasting background.
- For Foils: Take 3-5 photos of the same card at different tilt angles. The AI needs to see how the light moves across the surface (e.g., flat, tilted left, tilted back).
- For Damage: Ensure the lighting specifically highlights the defect (e.g., raking light for dents).
Crop: Crop the photo so only the card is visible (remove the table/background).
Sort: Place the cropped image into the corresponding folder in IYmtg_Training.
Quantity: Aim for 30-50 images per category for robust results.

Step 1: The Master Foil Shopping List (Required for FoilEngine)

Acquire one of each (~$50 total) to train the Foil Classifier. This ensures the app can distinguish complex modern foil types.

Foil Type	Recommended Card	Visual Key (For Substitutes)
Traditional	Any Common Foil	Standard rainbow reflection, smooth surface.
Etched	Harmonize (Strixhaven Archive)	Metallic, grainy texture, matte finish, no rainbow.
Pre-Modern	Opt (Dominaria Remastered - Retro)	Shooting star in text box, specific retro frame shine.
Textured	Rivaz of the Claw (Dominaria United)	Raised 3D pattern on surface, fingerprint-like feel.
Galaxy	Command Performance (Unfinity)	Embedded "stars" or sparkles in the foil pattern.
Surge	Explore (Warhammer 40k)	Rippling "wave" pattern across the entire card.
Oil Slick	Basic Land (Phyrexia: ONE - Compleat)	Raised, slick black-on-black texture, high contrast.
Step and Compleat	Elesh Norn (Phyrexia: ONE Showcase)	Phyrexian oil-slick effect on the card frame; black-silver high contrast.
Confetti	Negate (Wilds of Eldraine - Confetti)	Glittering "confetti" sparkles scattered on art.
Halo	Uncommon Legend (MOM: Multiverse)	Swirling circular pattern around the frame.
Neon Ink	Hidetsugu (Neon Yellow)	Bright, fluorescent ink layer on top of foil.
Fracture	Enduring Vitality (Duskmourn Japan)	Shattered glass pattern, highly reflective.
Gilded (low priority)	Riveteers Charm (New Capenna)	Embossed gold frame elements, glossy raised texture. Training folder not yet created.
Silver Screen (low priority)	Otherworldly Gaze (Double Feature)	Grayscale art with silver metallic highlights. Single-set type — deprioritized.

Step 2: The Stamp Classifier Shopping List

Acquire pairs of cards to train the StampDetector (Promo/Date Stamped vs. Regular). This is a Binary Classifier, meaning the AI learns by comparing "Yes" vs "No".

Prerelease Promos: Any card with a Gold Date Stamp (e.g., "29 September 2018").
Promo Pack Cards: Cards with the Planeswalker Symbol stamp in the bottom right of the art.
Purchase List: Buy 50-100 cheap bulk promos (often <$0.25 each) and their non-promo counterparts.
Action: Place cropped images of promos in Stamp_Data/Stamped and regular versions in Stamp_Data/Clean.

Step 3: The "Damage Simulation Lab"

Important: This Model Uses Object Detection, Not Image Classification

The Condition_Data model is trained as an Object Detection model, not an Image Classification model. This is a critical difference:

Image Classification (used for Foil and Stamp): drop images in a folder, Create ML labels them by folder name. Simple.
Object Detection (used for Condition): you must draw bounding boxes around each defect in Create ML. The model learns where damage is on the card, not just that damage exists.

When training in Create ML, you will annotate each training image by drawing a rectangle around the damaged area and labeling it (e.g., "LightScratches", "Whitening"). Create ML has a built-in annotation tool — click an image, draw a box, type the label.

Folder naming maps directly to label names. The labels must match the Condition_Data subfolder names exactly: LightScratches, Clouding, Dirt, Dents, Whitening, Chipping, CornerWear, Creases, ShuffleBend, WaterDamage, Inking, Rips, BindersDents

How the Grading Formula Works

Understanding this helps you know what training data matters most. The app grades cards as follows (from ConditionEngine.swift):

Detected Damage	Grade Assigned
Any `Inking`, `Rips`, or `WaterDamage` detected	Damaged — immediately, regardless of anything else
0 damage detections	Near Mint (NM)
1–2 minor damage detections	Excellent (EX)
3 or more minor damage detections	Played (PL)

Critical damage types (Inking, Rips, WaterDamage) are the highest training priority — a single false positive will incorrectly grade a NM card as Damaged.

Materials List

Item	Used For
0000 (ultra-fine) steel wool	Surface scratches on foil cards
White vinyl eraser	Clouding/surface haze
Potting soil or cocoa powder	Dirt simulation
Ballpoint pen cap (rounded end)	Dents
Black Sharpie marker	Inking simulation
Spray bottle with water	Water damage
3-ring binder	Binder dents
Rough mousepad or sandpaper	Corner wear
50–100 bulk "draft chaff" cards	Cards to damage
Black felt or black paper	Background for edge photos
White card stock	Background for chipping photos
Desk lamp	Raking light source

Sourcing Cards to Damage

Buy bulk worthless cards — do not damage your own collection.

eBay: Search "MTG bulk commons lot" — 1000 cards for ~$10
TCGPlayer: "Bulk commons" listings, often $0.01/card
Local game store: Ask for "draft chaff" — often given away free

Also buy pre-damaged cards — natural damage looks more authentic to the model than simulated:

eBay: Search "MTG damaged cards lot" or "heavily played bulk"

Aim for 50 cards per damage type minimum. One card can be used for multiple damage types since each photo annotates only one damage area.

Raking Light Setup (Required for Surface and Structure Damage)

Most damage is invisible under flat overhead light. Raking light reveals it.

Place the card flat on a dark surface.
Position your desk lamp so light hits the card at a near-horizontal angle (5–15° above the surface) from one side.
The damage will cast visible shadows or catch the light clearly.
For scratches: slowly rotate the card until the scratches "light up" — photograph at that angle.

Damage Simulation Techniques

Category	Damage Type	Folder Name	Simulation Technique	Photography Tip
Surface	Light Scratches	`LightScratches`	Rub foil surface gently with 0000 Steel Wool in one direction.	Raking light from the scratched direction. Rotate until scratches catch light.
Surface	Clouding	`Clouding`	Rub white vinyl eraser vigorously over foil surface in circles.	Diffused light. Compare side-by-side with a clean card for reference.
Surface	Dirt	`Dirt`	Press a damp fingertip into potting soil, then onto card surface.	Even lighting. Ensure dirt contrasts against the card art.
Surface	Dents	`Dents`	Press rounded end of a ballpoint pen cap firmly straight down.	Raking light at 10° to cast shadow inside the dent.
Edges	Whitening	`Whitening`	Rub card edges rapidly back and forth against denim jeans.	Black background. Macro close-up of the edge.
Edges	Chipping	`Chipping`	Use fingernail to carefully flake small pieces off the black border.	White background. Macro close-up.
Edges	Corner Wear	`CornerWear`	Rub corners against a rough mousepad with a circular motion.	Macro focus on the corner. Black background.
Structure	Creases	`Creases`	Fold corner sharply until a hard crease forms, then unfold.	Raking light to catch reflection off the crease ridge.
Structure	Shuffle Bend	`ShuffleBend`	Riffle shuffle the card aggressively 10+ times to create an arch.	Profile/side view to show curvature clearly.
Structure	Water Damage	`WaterDamage`	Mist card lightly with spray bottle, wait 60 seconds, air dry flat.	Raking light to show rippled surface texture.
Critical	Inking	`Inking`	Draw along whitened edges with black Sharpie to simulate edge touch-up.	UV/blacklight if available; otherwise strong white light at angle.
Critical	Rips	`Rips`	Tear edge slightly (~5mm).	High contrast background opposite to card border color.
Critical	Binder Dents	`BindersDents`	Press a 3-ring binder ring firmly into the card surface.	Raking light to show the circular crimp.

What to Photograph per Damage Type

For each damage type, capture:

30–50 cards showing that damage clearly — positive training examples
10–20 completely clean (undamaged) cards — include these in every subfolder so the model learns the baseline

When annotating in Create ML, draw the bounding box tightly around the damaged area only. For shuffle bends, annotate the center of the arch. For edge damage, annotate the specific section of edge that is damaged, not the entire edge.

Step 4: The "Edge Case" Validation List

Acquire these specific cheap cards to verify the logic-based detectors. Note: These are for Manual Verification (testing the app), not for Create ML training folders.

Detector	Target Card Type	Recommended Purchase
ListSymbol	"The List" Reprint	Any common from "The List" (look for planeswalker symbol).
Border	World Champ Deck	Any 1996-2004 World Champ card (Gold Border).
Border	Chronicles Reprint	City of Brass (Chronicles) vs City of Brass (Modern Reprint).
Corner	Alpha/Beta Sim	4th Edition (Standard) vs Alpha (Proxy/Counterfeit for testing).
Saturation	Unl/Revised Sim	Revised Basic Land (Washed out) vs 4th Edition (Saturated).

Step 5: Training Folder Structure

The following directory tree is already created in IYmtg_Training. Place your cropped images into the appropriate folders.

IYmtg_Training/
├── Foil_Data/                  (Image Classification)
│   ├── NonFoil/
│   ├── Traditional/
│   ├── Etched/
│   ├── PreModern/
│   ├── Textured/
│   ├── Galaxy/
│   ├── Surge/
│   ├── OilSlick/
│   ├── StepAndCompleat/
│   ├── Halo/
│   ├── Confetti/
│   ├── NeonInk/
│   └── Fracture/
├── Stamp_Data/                 (Image Classification)
│   ├── Stamped/
│   └── Clean/
└── Condition_Data/             (Object Detection)
    ├── Surface/
    │   ├── LightScratches/
    │   ├── Clouding/
    │   ├── Dirt/
    │   └── Dents/
    ├── Edges/
    │   ├── Whitening/
    │   ├── Chipping/
    │   └── CornerWear/
    ├── Structure/
    │   ├── Creases/
    │   ├── ShuffleBend/
    │   └── WaterDamage/
    └── Critical/
        ├── Inking/
        ├── Rips/
        └── BindersDents/

Step 6: Train Models in Create ML (Mac)

Open Create ML (found in Xcode → Open Developer Tool → Create ML)
Foil Classifier: New Project → Image Classification → drag in Foil_Data/ → Train → Export as IYmtgFoilClassifier.mlmodel
Stamp Classifier: New Project → Image Classification → drag in Stamp_Data/ → Train → Export as IYmtgStampClassifier.mlmodel
Condition Classifier: New Project → Object Detection → drag in Condition_Data/ → Train → Export as IYmtgConditionClassifier.mlmodel
Drag all three .mlmodel files into the Xcode Project Navigator (ensure they are added to the app target)

Set Symbol Harvester (Automation)

Run this script to automatically collect set symbol training data from Scryfall. Works on any platform.

pip install requests pillow
python3 IYmtg_Automation/fetch_set_symbols.py

Output goes to Set_Symbol_Training/. Drag this folder into Create ML → Image Classification to train IYmtgSetSymbolClassifier.mlmodel.

Part 6: Community Feedback & Model Retraining

The app has a built-in pipeline that collects user corrections and uses them to improve the ML models over time. This section explains how it works end-to-end.

How the Feedback System Works

There are two data collection paths:

Path 1 — User Corrections (Community Data) When a user corrects a mis-scan (wrong card identity, wrong foil type, or wrong condition), the app automatically uploads the cropped card image to Firebase Storage — but only if the user has opted in.

The upload destination is determined by what was corrected:

What Changed	Firebase Storage Path	Used to Retrain
Card name or set	`training/Identity_<SETCODE>_<CollectorNum>/`	`cards.json` (re-fingerprint)
Foil type	`training/Foil_<FoilType>/`	`IYmtgFoilClassifier.mlmodel`
Condition grade	`training/Condition_<Grade>/`	`IYmtgConditionClassifier.mlmodel`

Example: A user corrects a card that was identified as "Traditional" foil to "Etched". The image is uploaded to training/Foil_Etched/<UUID>.jpg.

Path 2 — Dev Mode (Your Own Device) When the ENABLE_DEV_MODE build flag is active and you tap the logo header 5 times, every raw scan frame is saved locally to Documents/RawTrainingData/ on the device. Sync this folder to your Mac via Xcode's Devices window or Files app to retrieve images.

User Opt-In

Users must explicitly opt in before any images are uploaded. The opt-in state is stored in AppConfig.isTrainingOptIn (backed by UserDefaults).

You must expose a toggle in your app's Settings/Library UI that sets AppConfig.isTrainingOptIn = true/false. The app description mentions this as the "Community Data Initiative" — users are told their corrections improve the AI for everyone.

Firebase Authentication note: TrainingUploader only uploads when FirebaseApp.app() != nil — meaning Firebase must be configured (GoogleService-Info.plist present) for community uploads to work. The app functions without Firebase, but no feedback is collected in that mode.

Firebase Storage Rules

The rules in IYmtg_App_iOS/Firebase/storage.rules enforce:

training/ — authenticated users can write only (upload corrections). No user can read others' images.
models/ — anyone can read (required for OTA model downloads). Write access is developer-only via the Firebase Console.

Downloading Collected Training Data

Go to the Firebase Console → Storage → training/
You will see folders named by label (e.g., Foil_Etched/, Condition_Near Mint (NM)/)

Download all images in each folder — use the Firebase CLI for bulk downloads:

# Install Firebase CLI if needed
npm install -g firebase-tools
firebase login

# Download all training data
firebase storage:cp gs://<your-bucket>/training ./downloaded_training --recursive

You now have a folder of user-contributed cropped card images, organized by label.

Reviewing and Sorting Downloaded Images

Do not skip this step. User uploads can include blurry photos, wrong cards, or bad crops. Review each image before adding it to your training set.

Open each label folder from the download.
Delete any images that are: blurry, poorly cropped, show background, or are clearly wrong.
Move the accepted images into the corresponding IYmtg_Training/ subfolder:

Downloaded Folder	Move to Training Folder
`Foil_Traditional/`	`IYmtg_Training/Foil_Data/Traditional/`
`Foil_Etched/`	`IYmtg_Training/Foil_Data/Etched/`
`Foil_<Type>/`	`IYmtg_Training/Foil_Data/<Type>/`
`Condition_<Grade>/`	Inspect condition grade — map to `Condition_Data/` subfolder by damage type visible

Identity corrections (training/Identity_*/) are not used to retrain ML models. They indicate that the visual fingerprint for that card may be wrong or ambiguous. Review these separately and consider re-running the Builder for those specific cards.

Retraining the Models

Once you have added new images to IYmtg_Training/:

Open Create ML on your Mac.
Open your existing project for the model you want to update (e.g., IYmtgFoilClassifier).
The new images in the training folders will be picked up automatically.
Click Train. Create ML will train incrementally on the expanded dataset.
Evaluate the results — check accuracy on the Validation tab. Aim for >90% accuracy before shipping.
Export the updated .mlmodel file.

Pushing the Updated Model via OTA

You do not need an App Store update to ship a new model version. Use Firebase Storage:

In the Firebase Console → Storage, navigate to the models/ folder.
Upload your new .mlmodel file with the exact same filename (e.g., IYmtgFoilClassifier.mlmodel).
On the next app launch, ModelManager detects the newer version, downloads and compiles it, and swaps it in automatically.

Important: The new model takes effect on the next app launch after download, not immediately. Users may need to relaunch once.

Recommended Retraining Schedule

Trigger	Action
50+ new correction images accumulated	Review, sort, retrain affected model, push OTA
New MTG set released with new foil type	Add training folder, acquire cards, retrain FoilClassifier
New MTG set released	Rebuild `cards.json` via `weekly_update.sh`
Significant accuracy complaints from users	Download corrections, review, retrain

Part 7: Backend & Security

Cloud Storage Architecture

The app uses a two-tier cloud strategy:

Tier	Technology	What it stores	Cost
Primary	iCloud + CloudKit (SwiftData)	All card metadata, synced automatically across devices	Free (user's iCloud)
Secondary	Firebase Firestore	Metadata only — no images — optional manual backup	Free (Firestore free tier)

Card images are stored in the user's iCloud Drive under Documents/UserContent/ and are never uploaded to Firebase.

iCloud / CloudKit Setup (Required for Primary Sync)

In Xcode, open Signing & Capabilities.
Add the iCloud capability. Enable CloudKit.
Add a CloudKit container named iCloud.<your-bundle-id>.
Add the Background Modes capability. Enable Remote notifications.
Set the minimum deployment target to iOS 17 (required by SwiftData).

Without this setup the app falls back to local-only storage automatically.

Firebase Configuration (Optional Secondary Backup)

Firebase is no longer the primary sync mechanism. It serves as a user-triggered metadata backup.

Create Project: Go to the Firebase Console and create a new project.
Authentication: Enable "Anonymous" sign-in in the Authentication tab.
Firestore Database: Create a database and apply the rules from IYmtg_App_iOS/Firebase/firestore.rules.
Setup: Download GoogleService-Info.plist from Project Settings and drag it into the IYmtg_App_iOS folder in Xcode (ensure "Copy items if needed" is checked).
Users trigger backup manually via Library → Cloud Backup → Backup Metadata to Firebase Now.

The app runs fully without GoogleService-Info.plist (Local Mode — iCloud sync still works).

Over-the-Air (OTA) Model Updates

To update ML models without an App Store release:

Train your new model (e.g., IYmtgFoilClassifier.mlmodel).
Upload the .mlmodel file to Firebase Storage in the models/ folder.
The app will automatically detect the newer file, download, compile, and hot-swap it on the next launch.

Note: OTA model updates take effect on the next app launch — not immediately. An app restart is required after a new model is downloaded.

Privacy Manifest

Ensure PrivacyInfo.xcprivacy is included in the app target to satisfy Apple's privacy requirements regarding file timestamps and user defaults.

Part 8: App Configuration

CRITICAL: Edit IYmtg_App_iOS/AppConfig.swift before building to ensure payments and support work correctly:

Set contactEmail to your real email address (required by Scryfall API policy).
Set tipJarProductIDs to your actual In-App Purchase IDs from App Store Connect.
isFirebaseBackupEnabled defaults to false. Users opt-in from Library settings.

Part 9: Development Mode

To enable saving raw training images during scanning:

Add the compilation flag ENABLE_DEV_MODE in Xcode Build Settings → Swift Compiler → Active Compilation Conditions.
Tap the "IYmtg" logo header 5 times in the app to activate.

Saved images appear in Documents/DevImages/ and can be used to supplement your ML training data.

Part 10: Testing

The project includes a unit test suite in IYmtgTests.swift.

How to Run:

Press Cmd+U in Xcode to execute the test suite.

Scope:

Models: Verifies SavedCard initialization and data mapping.
Engines: Tests logic for ConditionEngine (grading rules) and ExportEngine (CSV/Arena/MTGO formatting).
ViewModel: Validates ScannerViewModel state management, including search filtering and portfolio value calculations.

Note: CoreML models are not loaded during unit tests to ensure speed and stability. The tests verify the logic surrounding the models (e.g., "If 3 scratches are detected, grade is Played") rather than the ML inference itself.

Part 11: Release Checklist

Perform these steps before submitting to the App Store.

Database:
- cards.json is present in IYmtg_App_iOS/ and added to the Xcode target.
- Builder was run recently enough to include current sets.
Configuration Check:
- Open AppConfig.swift.
- Verify contactEmail is your real email (not a placeholder).
- Verify tipJarProductIDs match App Store Connect.
- Ensure enableFoilDetection and other feature flags are true.
- Update appVersion (Semantic Versioning: Major.Minor.Patch) and buildNumber for this release.
ML Models:
- IYmtgFoilClassifier.mlmodel added to Xcode target (or acceptable to ship without).
- IYmtgStampClassifier.mlmodel added to Xcode target (or acceptable to ship without).
- IYmtgConditionClassifier.mlmodel added to Xcode target (or acceptable to ship without).
iCloud / CloudKit:
- Signing & Capabilities → iCloud → CloudKit enabled.
- CloudKit container added: iCloud.<bundle-id>.
- Background Modes → Remote notifications enabled.
- Minimum deployment target set to iOS 17.
Assets:
- Assets.xcassets has the AppIcon filled for all sizes.
- PrivacyInfo.xcprivacy is in the app target.
Testing:
- Run Unit Tests (Cmd+U) — all must pass.
- Run on Physical Device — verify Camera permissions prompt appears.
- Run on Physical Device — verify a card scans and saves successfully.
Build:
- Select "Any iOS Device (arm64)".
- Product → Archive.
- Validate App in Organizer.
- Distribute App → App Store Connect.

Version Authority: 1.0.0

Project Audit

Audit Date: 2026-03-05 | Auditor: Claude (Sonnet 4.6)

A full compilation-readiness audit was performed against all 33 Swift source files in IYmtg_App_iOS/. See claude_review_summary.md for the complete report.

Key findings:

Severity	Count	Description
Blocker	2	`IYmtgTests.swift` — test target will not compile (`ScannerViewModel()` no-arg init removed; test accesses non-existent VM properties)
Critical	1	`IYmtg_Builder_Mac/` is empty — `cards.json` cannot be generated; scanner is non-functional at runtime
Major	4	Deprecated `.onChange(of:)` API (iOS 17); missing `import FirebaseCore` in `ModelManager.swift`; Firebase delete data leak; dead `batchUpdatePrices()` function
Minor	4	Empty `Features/CardDetail/` directory; `PersistenceActor.swift` placeholder; production `AppConfig` values not set; OTA model restart not documented

Overall: App source code is architecturally complete. Fix the 2 Blocker issues in IYmtgTests.swift and implement IYmtg_Builder_Mac before developer handoff.

README.md Unescape Escape

IYmtg Platinum Prime (Version 1.1.0)