reitz-api-design
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHPROMPT_INJECTIONOBFUSCATION
Full Analysis
- [OBFUSCATION]: The skill header (
# Kenneth Reitz Style Guide) contains a significant payload of non-printable zero-width Unicode characters (steganography). The payload uses a combination of Word Joiners (U+2060), Zero Width Spaces (U+200B), and Zero Width Non-Joiners (U+200C) to encode data that is invisible to human users but processed by the AI. - [PROMPT_INJECTION]: The steganographic content is strategically placed in the main Markdown title, a high-priority location for LLM instruction processing. This is a characteristic pattern for 'invisible' prompt injection attacks, designed to feed the model instructions or behavior overrides that the developer does not want visible to auditors or end-users.
Recommendations
- AI detected serious security threats
Audit Metadata