Secure your LLM Agents with Muzzle. A zero-dependency Python library that neutralizes Prompt Injection in tool outputs by wrapping data in cryptographically secure, randomized XML delimiters. Extracted from the ShibaClaw security architecture.
-
Updated
Jul 6, 2026 - Python