🃏 Classic1949WaddingtonsFIDDLYClue rebuild deck
FIDDLY
The board and the cards are an afternoon, and then one specific system will fight you: an AI opponent has to reason about what other players have shown each other, and a bot that only tracks its own cards is a bot that never solves the murder.
You will get it, but one specific system will fight you.
Sits at fiddly rather than easy purely because of the knowledge model. Ask a model for 'Clue with AI players' and you will get bots that track their own hand and guess — which looks fine for two turns and never solves anything. Ask for constraint propagation over a per-player candidate set and you get opponents that beat you, in about the same number of lines. The difference is entirely in knowing to ask.
- Difficulty
- FIDDLY · 2 of 5
- First playable
- A weekend
- The original
- 1 team-years
- Released
- 1949 · Classic
- Genre
- Deduction
Build a Clue-style deduction game with AI opponents that reason properly. Vanilla JS or React, hot-seat plus bots. Requirements: - Nine rooms, six suspects, six weapons. Deal one of each into the envelope, distribute the rest. Movement by dice roll or secret passage, suggestions made in the room you occupy, accusations end your game if wrong. Rename everything — the game is public domain in mechanic, not in trademark. - The knowledge model is the project, so build it first and build it as a constraint solver, not a checklist. For each player, maintain the set of cards they might hold. A player who cannot disprove a suggestion tells everyone at the table that they hold none of those three cards — that is a hard constraint. A player who does disprove it tells everyone they hold at least one, which is a disjunction. - Run constraint propagation after every suggestion, for the AI and for the human's notebook. When a disjunction collapses to a single card because of eliminations elsewhere, the notebook should update itself and say which suggestion caused it. - Auto-notebook UI: a grid of players against cards, marked has / has-not / unknown, filled in automatically from what was publicly shown. This is the feature that makes a digital version better than the cardboard one. - AI turns: move toward the room that eliminates the most possibilities, not the nearest one. Include the occasional suggestion naming a card the bot itself holds, to extract information without giving any. - Accusation confidence readout on a debug toggle, so you can see a bot close in. Definition of done: a bot makes a correct accusation from information that was all public, and the replay shows the three suggestions that got it there.
- Reading the person opposite you, which is where most of the real deduction happens at a table
- Bluffing suggestions to bait a reveal — bots do this badly unless you model it deliberately
- The trademarked characters, rooms and weapons; the deduction game underneath is yours, the Colonel is not
A notebook that fills itself in from public information
Every deduction game asks players to maintain a grid by hand, and most of what goes in that grid was announced out loud to everyone. Automate exactly the part that was public — never the part that was inferred — and you remove the bookkeeping while leaving the actual thinking untouched. Players stop losing to clerical error and start losing to the person who reasoned better, which is the game they thought they were playing.
ScopeA weekend— build just this, not the game
Build a deduction notebook as a single self-contained HTML file. DOM, no dependencies. - Four players, fifteen cards dealt out with three held back as the hidden solution. - Simulate a sequence of suggestions and responses. Each response is either 'cannot disprove' (a hard negative on three cards) or 'disproved' (a disjunction over three cards, with the specific card visible only to the asker). - Maintain for every player a candidate set, and run constraint propagation to fixpoint after each event. Collapse disjunctions when eliminations make them singular. - Render the grid live with three states, and highlight every cell the moment it is deduced, with a tooltip naming the event that forced it. - A step control to walk the event log forward and backward and watch the grid resolve. Definition of done: at least one cell in the grid resolves from a disjunction several turns after it was recorded, and the tooltip explains the chain that closed it.
Not a clone — Clue re-themed and cut down to something you can finish. The angle is the important reel: it is what turns a project into a weekend. 5,880 ways to land.
Obsidian One Room
Clue, but one level, done properly, set in a Mesoamerican empire. Rendered in claymation with baked light.
Build a browser game called "Obsidian One Room". The pitch in one line: a deduction in the spirit of Clue, but one level, done properly, set in a Mesoamerican empire. WHAT THIS IS Take the SHAPE of Clue — its core loop, and whatever makes that loop good — and rebuild it as your own game in a different world at a smaller scope. Use your own names, art and audio, which you want anyway: a reskin of somebody else's game is less interesting than a new one that works the same way. THE ANGLE — this is the important constraint, honour it above everything else One level, done properly. A single space, built to a finish quality you could not sustain across twenty. Every prop and every sightline is deliberate. This is what makes the project finishable. If a decision would push the scope back towards the original's, take the smaller option every time. THE THEME A Mesoamerican empire. Stepped stone architecture, an astronomical calendar that gates events, and obsidian-edged weapons that break. The theme is not a coat of paint. Let it change what the mechanics mean — a reload, a health pack and a locked door should all be things that make sense in this world and nowhere else. LOOK Claymation with baked light. Soft rounded geometry, a fingerprint-style normal texture, stop-motion timing at 12fps for characters while the camera stays at 60. REQUIREMENTS - Runs in a browser with no build step. Canvas 2D unless the angle genuinely requires 3D, in which case three.js. - Fixed-timestep update loop with interpolated rendering, so it behaves the same at 60Hz and 144Hz. - Every tuning constant in one CONFIG object at the top, exposed as live sliders in a debug panel. You will find the feel by dragging those, not by prompting. - All content — levels, entities, balance numbers — in JSON or plain data files, never inline in the logic. - Audio generated with the Web Audio API rather than asset files. - Respect prefers-reduced-motion: keep the fades, drop the shake and the parallax. BUILD ORDER 1. The core loop with placeholder rectangles, at the reduced scope the angle demands. No theme, no art, no audio. Make it fun as rectangles first. 2. The angle's consequences. Cutting to bots, or to turn-based, or to one level changes the design rather than just shrinking it — find out how and rewrite step 1. 3. The theme, as content and data. 4. The look, applied last as a rendering layer over a game that already works. Start by writing a one-page plan: the core loop, what the angle removes and what that frees you to do properly, and what you are deliberately leaving out. Wait for me to approve it before writing code.
Half the time one of these already does what you wanted, and the other half it saves you a week of solving a problem someone documented in 1997.
One count per game. No account — it is tied to a salted hash of your IP, which is not stored in a form anyone can reverse. Every build adds 1 team-years to the counter on the homepage.
Then try these
Questions
How hard is Clue to rebuild with an AI agent?
Yes — every game is buildable at some scope. The question is how hard, and Clue is FIDDLY (2 of 5). You will get it, but one specific system will fight you. The board and the cards are an afternoon, and then one specific system will fight you: an AI opponent has to reason about what other players have shown each other, and a bot that only tracks its own cards is a bot that never solves the murder. Reckon on a weekend to something you can actually play — not to something that matches the 1 team-years Waddingtons spent.
What makes Clue hard to rebuild?
The board and the cards are an afternoon, and then one specific system will fight you: an AI opponent has to reason about what other players have shown each other, and a bot that only tracks its own cards is a bot that never solves the murder. That is why it sits at FIDDLY rather than a tier either side of it. If you want the idea without the project, the transferable part is a notebook that fills itself in from public information — a weekend of work.
How long does it take to build a Clue clone?
Wrong question, and it is worth saying why. Nobody rebuilds Clue — the original took Waddingtons on the order of 1 team-years, released in 1949 on Classic. What you can do is reach something playable and genuinely yours in about a weekend, and then keep going for as long as it stays interesting. Any figure that claims otherwise is measuring a prototype and calling it a game.
What do I lose by building my own Clue?
Reading the person opposite you, which is where most of the real deduction happens at a table. Bluffing suggestions to bait a reveal — bots do this badly unless you model it deliberately. The trademarked characters, rooms and weapons; the deduction game underneath is yours, the Colonel is not. Those are the parts a prompt will not hand you.
Which AI coding agent should I use to build it?
Any of Claude Code, Codex or Cursor will handle this prompt. Paste it as your first message, let the agent scaffold the project, then iterate on feel — the second, third and tenth prompts are where a game actually gets good. Work in an empty folder or a fresh git branch so you can throw it away cheaply.
One email when new games go on the list. No other reason to email you, ever.