OIS Layer

Trustless omnichain intent settlement with agent competition.

OIS Layer

Created At

ETHGlobal Buenos Aires

Project Description

Omnichain Intent Settlement Layer (OIS Layer) is a universal cross-chain coordination network that allows users, dApps and autonomous agents to execute intent, any action a user wants to achieve, across multiple blockchains seamlessly, safely and trust-minimized.

OIS Layer introduces an execution marketplace, where specialized on-chain agents compete to fulfill user intents such as swaps, bridging, payments, NFT actions, gaming outcomes, DAO triggers or complex DeFi automations. Each intent is securely escrowed and evaluated and the best agent—based on price, speed, or reputation, wins the right to execute.

To guarantee secure omnichain execution, OIS Layer integrates:

  • LayerZero v2 for ultra-reliable, modular cross-chain messaging between chains like Ethereum, Base, Arbitrum, Optimism, Polygon, BSC, and Avalanche.

Chainlink CCIP for hardened, risk-managed cross-chain value transfer and contract calls.

Chainlink CRE (Computation Runtime Environment) to allow agents to run off-chain compute flows, perform optimization, routefinding, batching and verification.

Filecoin FVM for decentralized storage of agent proofs, execution traces, intent metadata and historical settlement logs.

This combination creates a verifiable, censorship-resistant, cross-chain intent settlement fabric where users define the outcome, and OIS finds the best execution path across the entire ecosystem—bridges, DEXs, automation networks, or agent engines.

The result is a unified omnichain infrastructure layer that unlocks next-gen DeFi, gaming, NFT flows, payments, liquidity routing and DAO automation, powered by trust-minimized cross-chain communication and decentralized compute.

How it's Made

How its made includes Ollama, Llama3.2, Chainlink, LayerZero, Filecoin, Cursor AI, Hardhat and custom smart contracts

The OIS Layer is built as a fully modular, omnichain intent-settlement engine powered by local LLM agents running on Ollama with Llama 3.2, allowing completely permissionless AI agents to compete, optimize, and execute user intents across multiple chains. These agents read contract state, run reasoning loops locally, generate execution strategies, and submit their proposed actions back to the blockchain.

On-chain logic is built using Solidity, deployed via Hardhat, and structured into several core modules:

IntentManager.sol for intent creation, validation, and lifecycle management

AgentRegistry.sol for agent staking & reputation

ExecutionProxy.sol for cross-chain execution routing

PaymentEscrow.sol for fee handling & conditional payments

ChainlinkOracleAdapter.sol for CCIP cross-chain messaging & CRE agent verification

For cross-chain execution, the system integrates LayerZero V2 OApp to route intents and execution packets between chains. The ExecutionProxy includes custom extensions beyond the base OApp logic so the project meets LayerZero’s requirement of innovating on top of the standard interfaces.

Chainlink CCIP is used as a secondary messaging layer and for settlement security, while Chainlink CRE validates that AI agents interacting with intents are auditable, cryptographically proven, and not fake/impersonated. The dual-messaging design increases execution reliability and decentralizes trust assumptions.

All agent reasoning metadata, execution logs, proofs, and snapshots of intent states are stored on Filecoin (via Web3.Storage/Lighthouse), enabling verifiable history, dispute resolution, and intent replaying.

The frontend is built with a minimal, animated architecture showcase (Cursor-generated) and communicates with the smart contracts through a modular SDK. Cursor is used throughout development to automate commits, track changes, and log implementation decisions (stored in an ignored “/dev-log” file). Testing relies on mock LayerZero endpoints, mock ERC20 tokens, and local Hardhat networks to fully simulate cross-chain flows before deploying to Sepolia Ethereum + Base Sepolia testnets.

Overall, the stack ties together local LLM intelligence, omnichain interoperability, secure agent verification, and decentralized storage — creating a trust-minimized, AI-powered intent settlement layer for any chain.

background image mobile

Join the mailing list

Get the latest news and updates