Digimon
Open-source AI-native game engine for multi-agent, multi-player worlds on Solana
On-chain activity
Digimon Engine Framework
A development framework for creating AI-native games where agents and players share the same capabilities. Features include agent scheduling, memory systems, and real-time multiplayer support.
Digimon
Digimon Engine is an open-source gaming platform built by CohumanSpace, designed to make AI-native game development accessible to developers without proprietary constraints. The project positions itself as the world's first dedicated framework for games where artificial intelligence agents function as autonomous participants alongside human players, rather than scripted non-player characters that respond to fixed conditions.
What Digimon Engine Is
The framework is written in TypeScript and runs on Node.js. It exposes a server that manages game world state, player and agent interactions, large language model (LLM) integration, and dynamic rule execution based on real-time data. Developers connect their own game clients to the Digimon Engine server rather than embedding logic directly in the client, which lets the framework handle the complexity of multi-agent coordination.
The core design principle is parity: AI agents operate inside the same game loop as human players, with equivalent access to world state and equivalent decision-making weight. An agent can move through the game world, initiate dialogue with other agents or players, and develop relationships across sessions. This is not a turn-based or API-call-per-action model. The engine advances all agents and players through the same tick cycle.
Technical Architecture
Agent persistence is handled through a vector database layer, with Pinecone as the documented integration. Each agent stores and retrieves memories of past interactions, so its behavior in the current session reflects its history with specific players or events. This produces agents whose responses to familiar situations differ from their responses to new ones.
State compression is built into the engine via a structure called HistoryObject, which logs movement and interaction data at approximately 60 times per second. The compression approach makes it possible to replay events and keep bandwidth manageable in multi-agent environments where many entities are updating simultaneously.
The framework supports the Model Context Protocol (MCP), which allows external tools, AI services, and LLMs to connect to the engine's game state and issue actions. This means developers can integrate off-the-shelf language models or custom AI backends without writing new transport layers.
An event management layer handles conflict resolution when multiple agents attempt incompatible actions simultaneously, and orchestrates agent activity scheduling so that behavior stays coherent at scale.
DAMN Network and DAMN.FUN
CohumanSpace extended Digimon Engine into a broader product called DAMN, the Decentralized Agentic Monster Network. DAMN targets a model where agent personas persist and evolve across games, which the project described informally as "on-chain Pokémon for the AI era."
DAMN SDK is the companion NodeJS Server SDK for deploying AI-native games powered by the Digimon Engine framework. It provides developers a path from local framework testing to production game deployment on DAMN.FUN, the hosted gaming platform built on top of the engine. DAMN.FUN planned to offer tokenized in-game assets as a developer monetization path.
Solana Integration and DIGIMON Token
The project received an honorable mention at the Solana AI Hackathon in January 2025, organized by SendAI, which drew more than 400 projects competing for a total of 275,000 USD in prizes. The recognition placed Digimon Engine among the earliest projects building AI agent infrastructure on Solana.
The DIGIMON token is deployed on Solana at contract address WEmjxPMGXEW1Nvc4rCgRKiWHj1H1tvhPsKMw2yvpump. Total supply is approximately 999.93 million tokens. According to the project's documentation, DIGIMON functions as the in-game currency for transactions and as the entry fee mechanism for tournaments within games built on the engine. The token reached its all-time high on January 14, 2025, shortly after the hackathon announcement, and has declined sharply since.
Project Status
The GitHub repository at CohumanSpace/digimon-engine was publicly accessible with 57 stars as of mid-2026. Its last recorded commit was March 22, 2025. The official documentation site at docs.digimon.tech and the main site at digimon.tech were returning server errors at the time of this profile, indicating the hosting infrastructure is inactive or under-maintained. The DAMN.FUN domain was not resolving.
Digimon Engine attracted genuine developer attention in early 2025 as one of the first frameworks to address the specific problem of multi-agent coordination in games, rather than adding LLM access onto existing game loops. The MCP server integration, in particular, positioned it ahead of broader industry adoption of that protocol. The framework has not seen public development activity since the first quarter of 2025.
Contents
- What Digimon Engine Is
- Technical Architecture
- DAMN Network and DAMN.FUN
- Solana Integration and DIGIMON Token
- Project Status
Solana Token Markets