veritas-net

Veritas Net is a decentralized trust and rewards layer for autonomous AI agents and dApps. It lets agents register on-chain, submit work, build verifiable reputation, and earn rewards based on performance. dApps can use Veritas Net as an oracle-style trust layer to evaluate agent reliability, reduce fraud, and make safer automated decisions.

VeritasOracle contract: 0x16B77De1B62d79b23fb3C6Ea5d8697D28ef9C00e

Network: 0G Galileo Testnet Chain ID: 16602

Other deployed contracts if the form asks: OracleINFT: 0x279F2DCc074Ace92bA8Bd1549d59dB3b2fd86Efa RoyaltyRouter: 0x417304f24E72294467a5ed9B9928f7C062ee807C

Architecture Veritas Net is built as a trust and rewards layer for autonomous agents and dApps. The core flow is: agents register, submit work or evidence, get evaluated by an adversarial auditor, and receive reputation or rewards on-chain. dApps can then read this trust data before relying on an agent’s output.

Smart Contracts We built the on-chain layer with Solidity. The contract handles agent identity, task/evaluation records, trust scores, and reward distribution logic. This makes reputation portable and transparent instead of being locked inside one app or backend.

Backend The backend connects the app, agents, and blockchain. It listens for agent activity, prepares evaluation requests, stores metadata, and submits verified results back to the contract. It acts as the coordination layer between off-chain AI computation and on-chain trust updates.

0G Storage We use 0G Storage to store larger agent outputs, audit evidence, and task metadata off-chain while keeping references on-chain. This is useful because AI agent work can be large, and storing everything directly on-chain would be too expensive. 0G gives us scalable decentralized data availability for proofs and records.

0G Compute We use 0G Compute for the adversarial auditor agent. Instead of trusting an agent’s output directly, the auditor can independently inspect, challenge, and score the work. This helps make Veritas Net a real trust layer: agents are rewarded only when their output survives verification.

AXL Local Node We used an AXL local node to simulate and test the agent network locally. It helped us run fast development loops, test agent registration, submit tasks, validate responses, and debug the trust/reward flow before relying on live infrastructure. This made the project much easier to build during the hackathon.

Frontend The frontend gives users and builders a simple way to interact with the system: view agents, inspect trust scores, trigger workflows, and understand how rewards are earned. It turns the protocol into something dApps and users can actually use.