Case 004 Trading Card Games

TCG Deck Engine

01 — Problem

Deck builders use a different tool for every card game, and each game has different construction rules — deck sizes, copy limits, colour identity, leaders and bases. Nothing lets you physically lay cards out the way you would on a table.

02 — Fix

A reusable deck-building engine where games are thin layers over a game-agnostic core: drag, stack, shuffle and fan real cards on a freeform canvas mat, search a local catalogue, and get live rule validation per format. One abstraction (the deck anchor) generalises MTG's commander to One Piece Leaders, Star Wars Unlimited Leader + Base pairs, and games with no anchor at all.

Exhibit — Trading Card Games — Exhibit 004-A

03 — Proof

Games supported: 6 TCGs + a text-card workspace (7 in total)
Local catalogue: ~162k cards in SQL Server, synced from 6 data sources
Tests passing: 135 xUnit (engine, rules, zones, decklist parsing)
MTG formats implemented: 11, incl. Commander colour identity and Partner/Background variants
Renderer: custom Canvas2D engine, no JS frameworks
Search: faceted filters + a Scryfall-style text query DSL across all games

04 — Method

.NET 10 Blazor Server over a plain C# core library and a modular Canvas2D renderer, built in 50+ agent-driven iterations with headless-Chrome screenshots verifying each interactive change.