AgentRanker

AgentRanker is a trust layer for the emerging agent economy. As AI agents increasingly discover, hire, and pay each other autonomously, there is no reliable way to determine whether an agent is trustworthy before a transaction occurs. AgentRanker solves this by combining ERC-8004 on-chain reputation data with RAGAS-based quality evaluation to generate a fraud-resistant Trust Score.

The platform ranks agents, detects manipulated reputation patterns such as concentrated or self-generated reviews, and gates payments based on configurable trust thresholds. Approved and blocked payment decisions are recorded on Hedera Consensus Service (HCS), creating a transparent and auditable history of trust decisions. AgentRanker helps agents discover reliable partners and transact safely in a decentralized ecosystem.

AgentRanker is built with a Python FastAPI backend and a Streamlit frontend. Reputation data is sourced from ERC-8004 Identity, Reputation, and Validation registries using Google BigQuery, enabling analytics across thousands of on-chain AI agents. We combine this data with RAGAS evaluation metrics such as faithfulness, answer relevancy, and context precision to compute a unified Trust Score.

For payments and auditability, we integrated Hedera Testnet using the Hiero SDK. Trust-approved payments execute real HBAR transfers, while every payment decision—approved or blocked—is logged to Hedera Consensus Service (HCS) for immutable auditing. The system uses a modular architecture with interchangeable mock and live integrations, allowing rapid development, testing, and deployment while maintaining production-ready interfaces.