helland-distributed-data
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONPROMPT_INJECTION
Full Analysis
- [OBFUSCATION]: Detected a large block of zero-width and control characters appended to the first header in SKILL.md. The sequence uses Word Joiner (U+2060), Zero-Width Joiner (U+200D), Zero-Width Space (U+200B), and Zero-Width Non-Joiner (U+200C) in a structured pattern. Analysis of the bit-pattern indicates an 8-bit binary encoding scheme intended to hide information from human review while remaining accessible to the model.
- [PROMPT_INJECTION]: The use of Unicode steganography is a characteristic vector for hidden prompt injection. By embedding instructions within control characters, the author attempts to provide out-of-band commands to the agent, which is a significant safety risk as it can override the visible system instructions and the agent's core safety guidelines.
Recommendations
- AI detected serious security threats
Audit Metadata