0xTrust.

Demo

About

The problem it 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/]

How it's made

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.