Skip to main content
Apply NowBuild your most ambitious dApp yet with Cartesi - grant applications open!

Escape from Tikal

Escape from Tikal

This is a cooperative boardgame in which the Player 2 is an LLM-powered AI. As we put the player 2 AI logic inside the Cartesi co-processor, players into a championship have transparency and verifiability on the fact that the player that cooperated with them didn't collude with another team to perform poorly and that every player has access to the same capable AI, ensuring fairness.
Hackathon ProjectGames

Team

Carlo Fragni
@carlofragni
Lyno Ferraz
@fergo6
Bruno Maia
@bruno.maia

Share project

Gallery

About Escape from Tikal

We developed a deterministic game engine in Python that is used both to execute the game while it's being played (inside an offchain backend) and verified (inside the Co-processor Cartesi Machine). The flow starts with a challenge being created in the DApp smart contract, locking a prize that can be claimed by the player with the highest score after the challenge period ends. After that, the challenge it is available in the front end to be played. Selecting it requests the game map to the Python offchain backend, which generates it in a deterministic fashion using the challenge ID as a seed. The player then writes a word in the form available in the frontend and it's sent to the offchain backend, where the game engine calculates all tiles where it can be placed along with the neighboring words of every valid tile. The word groups are sent to the Phi3-mini-4k LLM with a prompt requesting which word group is the closest related to the provided word. Once the LLM provides a response, the game engine moves to a tile neighbouring the word group chosen by the LLM (if there is more than one option, it's chosen based on a random value generated using the board state as a deterministic seed) and evaluates the game rules according to the content of the tile. The player chooses a new word and that goes on until either they escape the maze or die. Once the game is completed, the words played are submitted onchain along with the challenge ID, and the Co-processor uses the same game engine to verify the match and submit the score onchain. After the challenge period is over the player with the highest score can withdraw the prize associated with the challenge.

What's next

Improve the fontend (specially aesthetics), create new game modes and tile kinds.Add support to other languages.

Project founded on: Feb 18, 2025
Anyone is free to submit information about their project. Do your own research and use your best judgment when using or interacting with any of the projects listed in this directory. Being listed in this directory is not an endorsement from the Cartesi Foundation or any other related entity.

Explore similar projects

Cartesi Super Soccer
Hackathon Project

Cartesi Super Soccer

Games

Cartesi Soccer Manager lets players build NFT teams, send lineups to the blockchain, and earn XP after each match. The Co-Processor calculates results and updates NFT levels almost instantly.

Project founded on: Feb 18, 2025
ThinkChain
Hackathon Project

ThinkChain

AI/ML

On-chain service that enables smart contracts to perform verifiable large language model (LLM) inference.

Project founded on: Feb 18, 2025
TrustMatch
Proof of concept

TrustMatch

Games

Matchmaking in e-sports is the process of pairing players in teams against each other in a fair and competitive manner. Matchmaking systems are crucial in ranked and casual play, ensuring that games are challenging yet enjoyable. So, if done in a centralized manner, the matches can be easily biased, and that's why we propose this project, a matchmaking system using Cartesi Coprocessor.

Project founded on: Feb 18, 2025