HN Explorer

aunicall

↗ HN profile [ 18.5 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. Why I'm moving away from Regex for LLM Agent security

    I’ve been auditing how open-source execution engines handle prompt injection. Most of them (like OpenClaw) rely on a 3-layer static defense: regex blacklists, XML tagging, and character sanitization. The problem is that…

    3 points 3
  2. Analyzing OpenClaw's 3-layer defense against prompt injection

    I’ve been analyzing how open-source AI execution engines (like OpenClaw) handle prompt injection. The conclusion is concerning: when an Agent has tool access (shell, DB, web), Prompt Injection is no longer just…

    1 points 1