ClawTrap - MITM-Based Red-Teaming for OpenClaw Security Evaluation

This research introduces ClawTrap, a MITM-based red-teaming framework designed for the real-world security evaluation of autonomous web agents like OpenClaw. To bridge the gap between static sandbox testing and live network threats, ClawTrap provides a reproducible pipeline for rule-driven interception, transformation, and auditing at the network layer.