muratori-performance-aware
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONPROMPT_INJECTION
Full Analysis
- [OBFUSCATION]: The primary header in the SKILL.md file contains a sequence of zero-width Unicode characters (U+200B and U+200C) appended to the title text. This steganographic technique encodes data that is invisible to the user in standard markdown renders but can be processed by the agent.
- [PROMPT_INJECTION]: The presence of hidden Unicode data in the document header is identified as a steganographic prompt injection vector. This technique is designed to inject instructions into the AI context that are not visible to human reviewers, which can be leveraged to override safety constraints or operational rules.
Recommendations
- AI detected serious security threats
Audit Metadata