Invisible Unicode And Encoding Based Obfuscation Detection

via “invisible unicode and encoding-based obfuscation detection”

Open-source LLM input/output security scanner toolkit.

Unique: Specialized detection for unicode-based obfuscation techniques (zero-width characters, homoglyphs, combining marks) that other scanners may miss; analyzes character encodings at the unicode level rather than semantic level; prevents evasion of other security scanners through encoding tricks

vs others: More targeted than generic text sanitization because it specifically detects obfuscation patterns; complements other scanners by catching evasion attempts that use unicode tricks; runs locally with no external dependencies