AI Security | Shaoyang Cuihttps://spidermonk7.github.io/tags/ai-security/AI SecurityHugo Blox Builder (https://hugoblox.com)en-usThu, 02 Apr 2026 00:00:00 +0000https://spidermonk7.github.io/media/icon_hu7729264130191091259.pngAI Securityhttps://spidermonk7.github.io/tags/ai-security/ClawTrap - MITM-Based Red-Teaming for OpenClaw Security Evaluationhttps://spidermonk7.github.io/ongoing-projects/clawtrap/Thu, 02 Apr 2026 00:00:00 +0000https://spidermonk7.github.io/ongoing-projects/clawtrap/<p>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.</p>