turon-api-design
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONPROMPT_INJECTION
Full Analysis
- [OBFUSCATION]: The skill uses zero-width Unicode characters (U+200B, U+200C, U+200D, and U+2060) to hide a binary-encoded payload within the main title of the skill. This is a form of steganography designed to conceal content from human users while remaining readable by the AI agent.
- Evidence: The title
# Aaron Turon Style GuideinSKILL.mdcontains the hidden sequence:. - [PROMPT_INJECTION]: The use of hidden Unicode characters in a system instruction context is a high-risk indicator of prompt injection. The concealed payload is intended to provide instructions to the AI that differ from the visible text, potentially overriding agent behavior or bypassing safety constraints.
- Evidence: The presence of a complex, bit-encoded hidden string at the very beginning of the instruction file (within the header) suggests an attempt to establish malicious context before legitimate instructions are processed.
Recommendations
- AI detected serious security threats
Audit Metadata