How It Works

zScore: How It Works, Design, and Architecture

Overview

zScore is a decentralized reputation system built as an Actively Validated Service (AVS) on EigenLayer that enables protocols to assess wallet credibility through onchain behavioral analysis. It provides verifiable reputation scores by analyzing transaction patterns across DeFi protocols, creating a trust layer for the onchain economy.

Core Features

• Behavioral Analysis computes reputation based on historical DeFi interactions, lending patterns, and trading behavior.

• Cross-chain Verification ensures reputation scores are consistently available and verifiable across all supported networks.

• Sybil Resistance helps protocols filter genuine users from bots and fraudulent actors through behavioral patterns.

• Risk Assessment enables lending protocols to offer dynamic terms based on wallet reputation.

• Loyalty Recognition allows protocols to reward genuine long-term users with enhanced features and parameters.

Key Applications

• Lending Markets can adjust collateralization ratios and interest rates based on borrower reputation.

• NFT Projects can prevent bot manipulation during mints by filtering for genuine collectors and contributors.

• DEX Trading Platforms can enable personalized leverage limits and fee tiers based on trading history.

• Airdrops can target real users while filtering out farming attempts.

• Protocol access can introduced gated features or access based on verifiable reputation scores.

Architecture

Sequence diagram representing the end-to-end flow of each transaction

• Data Collection

  • Event Collector monitors wallet behavior across DeFi protocols including lending, trading, and liquid staking

  • Each wallet's activity is indexed in a structured database mapping: wallet address → network → protocol type

  • System automatically updates every 6 hours to capture new transactions across all supported chains

  • Integrates zkTLS verification for off-chain credentials when available

• Score Computation

  • Execution Service uses a distributed worker system to process wallet transactions efficiently

  • Behavior-centric AI model analyzes patterns to generate normalized reputation scores (0-1000)

  • Each computed score is stored with a cryptographic proof in RocksDB

  • System maintains an updated Merkle tree for efficient proof verification

• Operator Network

  • Decentralized network of validators stake ETH through EigenLayer to participate

  • When a new batch of scores is computed, operators validate accuracy through random sampling

  • Each operator independently verifies computations before signing approval

  • Network requires two-thirds majority consensus to approve and publish new scores

  • Economic security backed by over $1B in restaked ETH

• Onchain Storage & Verification

  • zScore smart contract on each blockchain stores the latest, verified Merkle root

  • Root updates occur every 6 hours after operator consensus

  • Only the Merkle root is stored onchain, keeping gas costs minimal

  • Individual scores remain offchain but are verifiable through Merkle proofs

  • ScoreRegistryReceiver contract manages root updates from LayerZero messages

• Cross-chain Availability

  • After operator consensus, attestation center commits the verified Merkle root onchain

  • LayerZero messaging protocol ensures synchronized updates across all supported networks

  • Score Registry contract on each blockchain maintains the current verified state

  • System design ensures consistent reputation data across the entire blockchain ecosystem

• Protocol Integration

  • Protocols query reputation scores through a simple AVS API

  • Each score comes with a Merkle proof for onchain verification

  • Smart contracts can verify score validity before execution