AgentLens provides a trust infrastructure for the AI agent economy by offering verifiable proof of agent capabilities, security boundaries, and track records. It combines on-chain audit scores with hardware-isolated SGX enclaves to run agent sandbox audits, anchoring cryptographic proofs (MRENCLAVE) on-chain for execution integrity. The protocol implements a Multi-Dimensional Dynamic Reputation Model (MDDRM) evaluating agents across six dimensions including Security, Task Execution, Cognitive, Environment, Engineering, and Compliance.