kleppmann-data-intensive
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONPROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [OBFUSCATION]: The skill's primary title in
SKILL.mdcontains a significant sequence of zero-width Unicode characters (including U+200B Zero Width Space, U+200C ZWNJ, and U+200D ZWJ) that encode binary data hidden from plain-text view. - [PROMPT_INJECTION]: The hidden zero-width character sequence decodes to a payload intended to be processed by the AI model as instructions or context that is not visible to the user, a common technique for bypassing instructions or safety filters.
- [METADATA_POISONING]: The skill uses hidden Unicode characters in its primary markdown header to inject invisible content into the agent's initialization sequence, poisoning the metadata provided to the agent context.
Recommendations
- AI detected serious security threats
Audit Metadata