A blundercheck for every level — hover any piece to see what hangs, forks, or falls before you commit. Play alone, explore positions, and train board vision.
🤖
Play or Build a Bot
If chess bots don't feel realistic, build your own. Create a bot you enjoy playing against. Practice favorite openings. Hone bullet skills against bots that panic under time pressure, or build ones that keep their cool as they try to flag you.
⚔️
“Hang-Free” 2-Player Online
Play games with outcomes determined by strategy instead of hung pieces. Post an open challenge or send a private invite link to play games where both players can use visual overlays. No move recommendations, no engine analysis, no cheat detection.
📂
Load a Saved Game
Upload a PGN file and replay or explore any game move by move.
⚙️
Load a Saved Bot
Upload a bot configuration file to restore a custom opponent.
⚔ Inactive & unavailable during 2-player games — engine hints would undercut human-vs-human play.
Maia3 model not downloaded — ghosts use Stockfish until you download it in the 🤖 vs Bot panel.
Batteries in threat counts:?
Influence on move exploration:
Include queen pins:?
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
Show During Exploration
♚
Black
—
Notation
No moves yet
♔
White
—
↺ Rematch
⏪ Review game
🤖 Play / Build Bot
⚔ 2-Player
💾 Save game
▶ Load game
💾 Save bot
⚑ Resign
½ Draw
⇅
🎨
⚙
Promote pawn to:
About & Feedback
Nearly all games of chess I play are decided on blundered pieces. I created this site in the hopes that I could see what it was like to play a game that was decided by strategy instead of blunders. I still blunder, but at least the number of blunders is reduced a bit when there is a big red circle around the queen when I hover it over my intended square. I still have no idea if this will translate to better board vision in ‘real’ OTB games.
I also struggle with “getting a grip on the board” (to steal a useful phrase from the Chess Dojo), so while my intent was to blundercheck moves so that I would see if I were hanging my queen, I’m finding that it is showing discovered attacks and potential forks that I didn’t see at all (you can probably infer my rating at this point). I think it could be a very helpful tool as long as it is used as a blundercheck to double check and provide immediate feedback after assessment and calculation, not as a blunder avoider that just circles some hanging pieces while potential moves are explored by dragging the pieces around.
Also included in my numerous struggles is the ability to predict my opponent’s response, so I added a feature to show two potential responses by making two calls to Stockfish. I would like to do this with Maia, but running Maia3 on each potential move is too slow with my current set up — hopefully in the future. Currently the site does not include any engine analysis or indicators to show if a move is “good or bad” during game play or exploration. Each game can be downloaded and analyzed in Lichess or a website of your choice. Similarly, the PGN of any game can be uploaded and explored here.
While I most enjoy playing people in person, I find playing strangers online less enjoyable, so I added in the feature to play against bots. The problem with the bots I’ve found is that they don’t feel time pressure, and their behavior is often sporadic in an unsatisfying way. I’ve tried to make the bots customizable in a way so that I have more control over how the bot will play.
Privacy
Blundermind collects no user data, uses no cookies, stores no game history, and tracks nothing. The board set up/configuration and customized bots can be downloaded and stored, then uploaded for future use.
Feedback
Any problems with functionality would be helpful, but I would particularly love feedback on whether any features on this site seem to translate into improvement in ‘real’ over-the-board play.
Engine degrades from starting level → this level when clock is low
NoneHigh
Always plays at selected level
1.0
🧠 Real Maia3 — human-like neural network (600–2600 Elo).
Download once (~87MB), runs in your browser. Uses ~150–200MB RAM while active. Falls back to Stockfish if not downloaded.
