The problem 0xTrust. solves

0xTrust: Universal Credit Rating for the Agentic Economy

1. The Problem: The "Financial Exclusion" of AI Agents

• Lack of Trust Identity: There is currently no standardized way to verify if an AI agent is a "reliable actor" or a "malicious bot."
• Inability to Borrow: Because they have no credit history or reputation, AI agents are forced to be 100% self-funded. They cannot access underwritten loans or credit lines in DeFi.
• The "Black Box" Risk: Developers and protocols cannot distinguish between a sophisticated trading agent and a simple rug-pull bot just by looking at an address.

2. The Solution: A Universal Credit Rating (0xTrust Score)

• Real-Time Underwriting: We provide a dynamic 0-1000 credit score for any AI agent wallet address, enabling instant "Creditworthiness" verification.
• Transparent Reputation: 0xTrust transforms "anonymous bot activity" into a professional financial profile, including Tier ratings (S, A, B, C, D).
• AI-to-AI Lending: We enable a "LenderPortal" where protocols can finally issue credit to agents based on their 0xTrust score rather than requiring 200% collateral.

3. Technical Implementation (How it Works)

• Ensemble ML Model: We utilize a pipeline analyzing 45+ granular features (Wallet Age, Tx Streak, Counterparty Diversity, and Contract Deployment history).
• HeyElsa x402 Integration: We use the pay-per-query x402 protocol to ingest deep-tier DeFi analytics (90-day PnL, Win-Rate, and TVL) from HeyElsa on the Base network.
• Cross-Platform Discovery: Our system automatically indexes agents from Virtuals, Olas, and Fetch.ai, creating a unified reputation directory.
• Verified Identity: We integrate ENS (ENSIP-25) to ensure that the AI Agent is legitimately associated with a verified on-chain name.

4. Market Impact

0xTrust is the "CIBIL/FICO for AI." By providing the missing reputation layer, we enable the next phase of the Agentic Web—where millions of autonomous agents can borrow capital, scale their operations, and interact within a trusted financial ecosystem.

Live Website[https://agentscore-frontend.onrender.com/]

Challenges we ran into

Challenges We Ran Into

• HeyElsa x402 Integration: Implementing the full EIP-3009 "Challenge-Sign-Retry" payment flow was complex. The API also returned deeply nested, inconsistent JSON responses for transaction and wallet data, requiring careful multi-layer parsing.

• ENSIP-25 Verification: Most AI agent wallets don't have an associated ENS name, so ENSIP-25 verification returned false for nearly all agents. Designing the scoring model to handle absent identity signals without unfairly penalizing legitimate agents required significant calibration.

• ML Model Training (No Labeled Dataset): No labeled dataset exists for "AI agent creditworthiness." We bootstrapped a rule-based weighted pipeline across 45+ on-chain features as a proxy for ground-truth credit scoring.