Hyperion Hackathon
$200K in Rewards
Claim Layer3 rewards
%20(1).png)
As part of the Hyperhack developer program, we hosted a live workshop introducing the Hyperion Testnet DEX—a Uniswap V2-style decentralized exchange deployed on the Hyperion Layer 2 network. The goal of the session was to demonstrate how developers can interact with this DEX, build AI-driven trading agents, and leverage Hyperion’s real-time architecture for intelligent DeFi workflows.
This blog recaps the workshop, outlines the tools available, and explains how to integrate AI agents into trading applications on Hyperion.
The Hyperion Testnet DEX is designed to provide developers with a single unified environment for trading-related activity within the Hyperhack. Built to simulate production-like interactions, it allows participants to explore how their agents and tools can operate in real time.
Key goals of the DEX include:
You can find documentation and developer resources at Hyperion Doc.
The DEX is accessible at Test DEX. Upon accessing the interface, users can interact with familiar DeFi functions including:
Supported token pairs currently include:
Although this is a testnet deployment, it mirrors Uniswap V2 mechanics, offering developers a realistic testing environment. It’s worth noting that the absence of trading bots means price fluctuations can persist longer than in live markets, making it ideal for simulating arbitrage or strategy-driven trading scenarios.
If you need test tokens to begin interacting with the DEX, you can request them via the Hyperhack Forum.
To enable data-driven applications, the DEX is backed by a GraphQL subgraph that indexes:
This subgraph allows developers and AI agents to perform granular queries, such as:
This infrastructure enables deeper integration with AI agents that need to reason over market state or build predictive models.
To streamline agent development, we’ve deployed an MCP (Meta Control Protocol) server. The MCP server acts as a lightweight RESTful wrapper over the subgraph, abstracting common queries into accessible API endpoints.
Functionality includes:
The MCP was purpose-built to simplify integration with external services, including AI toolkits and automation platforms. It is particularly effective for use with AI agents developed using the Alith SDK or similar frameworks.
During the session, we outlined how to integrate Hyperion DEX into an AI agent workflow. The architecture includes:
The result is a modular and composable AI-driven trading system capable of making autonomous decisions based on real-time DEX data.
We showcased an example involving Alith agents, where:
A question was raised during the session regarding how AI is integrated into the Hyperion ecosystem.
While full model inference onchain remains limited by current infrastructure, Hyperion allows for AI-native architectural design, where reasoning and decision-making workflows leverage onchain data and logic.
For the purposes of the Hyperhack:
AI inferencing can be done offchain while referencing onchain data indexed by the subgraph or fetched through the MCP server. This hybrid approach allows for powerful AI use cases without compromising scalability or cost.
The Hyperhack Ideathon is open until June 15, 2025. There’s still time to apply if you are working on an idea or want to build on Hyperion.
Tracks include:
Participants are encouraged to submit projects that demonstrate:
The Hyperion Testnet DEX and supporting infrastructure—including the subgraph and MCP server—provide a robust foundation for building and testing AI-powered DeFi agents. By using these tools, developers can explore cutting-edge use cases such as autonomous trading, liquidity automation, and intelligent analysis on a performant Layer 2 network.
We look forward to seeing the innovations that emerge from the Hyperhack and invite all developers to participate, ask questions, and build the future of decentralized intelligence.
