You have a mistake in Game protocol …
Next line: snakebotCount, one integer for the number of remaining snakebots on the grid.
Next powerSourceCount lines: The following 2 inputs for each power source:
3 Likes
Snake could go out of board at the top:
https://www.codingame.com/replay/875601358
Hello Everyone. I have 2 questions:
- Does the official referee treat movement into a just-vacated body cell as a collision or not?
- Are bottom-only out-of-bounds removals enough, or does the referee remove snakebots for any off-grid fall state?
Moving into a cell another snake’s tail just left is not a collision: all snakes move first, then collision checks use the updated bodies.
Snakes die when they fall out the bottom, not when they leave the grid on the sides or top. If it looks like they die as soon as they go off-grid anywhere, it’s because leaving the sides or top makes them fall, and they hit the bottom death condition shortly after.
1 Like
Am I missing something or are the snakes blind? I cannot figure out how to tell where walls or even my head is for the snakes. I see we have body length and coordinates for power sources, but no coordinates for the snakes or walls.
It’s a bird ? It’s a plane ? No, it’s a snake !
If you look at the github code they provided, they actually never refer to the snakes as something else than a bird.
Yup.
Have a look at the game protocol at the end of the puzzle input.
- Walls are given once before the game loop.
- Snakes coordinates are given on each turn in the game loop, their head is the first coordinates in the body variable.
Cheers.
1 Like
If you look at the github code they provided, they actually never refer to the snakes as something else than a bird.
Trying to spot the people using A.I : - )
Hi,
All my games (both in editor and in Arena) look like this horrifying 3-player tangled snakes game:
Am I really unlucky with all my games or is there maybe a bug in the visualizer/runner?
Cheers,
1 Like
Found one of those too, here’s a replay https://www.codingame.com/replay/876365117
Snakebod 0 seems to be discontinuous, here the body parts in order: 0 [[50, 11], [51, 11], [52, 8], [52, 9], [52, 10], [52, 11], [53, 8], [54, 8], [55, 8], [56, 8], [56, 9], [57, 9], [58, 9], [59, 9], [60, 8], [60, 9]]
Other snakebots are also discontinuous
2 Likes
Impressive replay.
I guess you’ve found a super bug.
Never face this problem but need to play more I guess to face it.
Well done for this discovery anyway!
In fact because of the gravity reason,
it’s not a snake game,
it’s more like a fish-bird so in fact it was really a bird at the build of the code.
The snake design is the packaging or the game design to give the illusion it’s an easy game like a snake, but it’s a new game fish-bird game!
Running the replay viewer locally and view in browser
Hi, I am trying to run the provided game engine from github locally using
$ mvn compile exec:java -Dexec.mainClass=Main -Dexec.classpathScope=test
but http://localhost:8888/test.html gives me 404.
What am i missing?
Try the brutaltester compatible version GitHub - aexg/WinterChallenge2026-Exotec: CodinGame Challange · GitHub
It looks like you cannot place markers with negative coordinates even though the snakes can go there. You can, however, place them outside of the map on the other side
Hi everyone,
I’m currently at 10.98 and trying to reach 50+.
My bot already has:
-
exact referee simulation (move → eat → behead → fall)
-
maximin search vs opponent action candidates
-
anti-stall memory (penalize repeated no-growth moves)
-
per-snake apple targeting + space-control scoring
Current problem: in some matches, my snakes still fall into repetitive movement mid-game and lose apple tempo.
If you have 2-3 high-impact tips for Gold/Legend (search budget, eval weights, opening plan, or anti-trap logic), I’d really appreciate it.
I can share replay IDs and code snippets if needed. Thanks!
- Path finding (without full simulation) seems important. You can use a modified version of Djikstra to include gravity & allows some repetitive loops that change the snake body to help it climb. You will know then which foods are unreachable for which snakes.
- Make sure the snakes go to different foods.
- I ignore moves that go outside the grid to do full simulations over fixed memory.
1 Like
Hi all !
I’m facing Error (#407) : “You reached the limit of plays for a period of time.” this afternoon. I wasn’t aware of limitations and now I’m stuck 
Could you please kindly tell how many plays we are allowed on which period of time? Or is the information available somewhere I haven’t seen?
Thanks a lot for your help!
EDIT: The error lasted just a few minutes, but I’d be curious still to learn more about the limitations.
Hi everyone. How is the winner chosen when there is a tie ? At first, I thought that the last one to eat a power source loses (it had happen with the boss). But in the following replay, the opponent ate the power source last and still won. Thanks in advance if you have any information ! 
In the shared replay I see a tie with same points and no body parts losses
each player is ranked “1st” on the final screen, and each one has a “star” under the viewer
there is no “winner” to declare
1 Like