Built in public.
TCGON is a personal project. It began as a simulator for a single title and is being redesigned through live testing with a second TCG, separating title-specific behavior from the shared engine. This page records that broad arc.
Timeline
- Card Game Simulatordone
Started out as a card game simulator for a specific title.
- Generic Game Enginein progress
Using a second TCG to uncover title-specific assumptions and move game-specific decisions into definitions.
- Data-driven Rulesdone
Moved to defining boards, zones, phases and rules as data.
- TEDLin progress
Developing a common representation that translates card text into triggers, costs, targets and actions.
- On-demand Translationin progress
Translating card titles and effect text on demand while preserving the source data.
- TCGON Builderin progress
Building out the environment for authoring game data.
- AI Playerin progress
Verifying an AI that selects from the actions the engine offers.
- AI vs AIin progress
Developing AI-vs-AI self-play through the same command pipeline.
- Multi-TCG Experimentsin progress
Testing compatibility of TCGs with different rule systems on one engine.
Development log
Detective Conan Card Game development history
Parts 1–4 document the experiments and setbacks of the project's single-title simulator era.
- From Unity to React: Building a Detective Conan Card Game Simulator in the Browser | Development Story 1The move from Unity to the browser, beginning with a deck editor and network-ready architecture.Read on note ↗
- Why Moving Cards Was Harder Than Expected | Detective Conan Card Game Development Story 2Implementing card movement, rotation and stacking while confronting the challenges of publishing a fan app.Read on note ↗
- A Year of Building and Rebuilding a Detective Conan Card Game Simulator | Development Story 3A year of changing specifications, rebuilding features and assembling the systems needed to run a game.Read on note ↗
- A Year In, Still Unable to Publish — Yet Development Continued | Detective Conan Card Game Development Story 4Defining card effects led to two new goals: AI matches and a generic card game engine.Read on note ↗
TCGON — the shift to a generic engine
From Part 5 onward, the log follows TCGON as it moves toward one engine for multiple TCGs, humans and AI.
- Leaving a Single-Game Simulator Behind: The Beginning of TCGON | Development Story 5The project leaves its single-title implementation behind and takes its first step toward running multiple TCGs on TCGON.Read on note ↗
- Humans No Longer Required: Running AI-vs-AI Matches on TCGON | Development Story 6AI players compete through the same command pipeline as humans, creating a foundation for automated game testing.Read on note ↗
- Teaching AI to Play Cards Became a TCG Development Environment | Development Story 7 — FinaleThe finale adds generic AI, decision logs and analysis, completing the shift toward an environment for building and testing TCGs.Read on note ↗
- How the Generic TCG Simulator TCGON Is Built | Architecture OverviewA technical overview of the Builder, runtime data, game engine, command pipeline and shared human/AI architecture.Read on note ↗
- From Generic in Theory to Proven in Practice: Running ONE PIECE CARD GAME on TCGON | Development Log 1A second TCG puts TCGON's generic engine to the test, reusing its board, setup, turn and card-play systems to reach a basic solo-playable state.Read on note ↗
- Trying to Build ONE PIECE Card Effects, I Rebuilt TCGON Itself | TCGON Development Log 2Battle flow and rendering are rebuilt as card effects begin to move toward a shared cross-title model.Read on note ↗
- Building ONE PIECE Cards One by One Ended with ‘Do the Rest’ | TCGON Development Log 3The unit of work shifts from cards to effect patterns, using shared parts and classification at scale.Read on note ↗
- Asking for Every ONE PIECE Card Effect Led to a Translation System | TCGON Development Log 4Per-card implementation gives way to a system that translates card text into shared meaning.Read on note ↗
- Translating Every Remaining ONE PIECE Card Effect Was Only the End of the Beginning | TCGON Development Log 5The vocab pipeline classifies and translates 3,694 effects, turns missing behavior into shared engine capabilities, and moves 2,823 card types into formal implementation before live-game testing.Read on note ↗
- ONE PIECE Revealed That TCGON Was Only Generic in Appearance | TCGON Development Log 6Live play exposes hidden Detective Conan assumptions, prompting a refactor that treats Conan and ONE PIECE as equal game definitions on the same engine.Read on note ↗