alexandrescu-modern-cpp-design
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONPROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [OBFUSCATION]: The skill title contains 176 hidden Unicode characters (Zero-Width Space
U+200B, Zero-Width Non-JoinerU+200C, and Zero-Width JoinerU+200D). This steganographic technique is used to embed data that is invisible to human readers but processed by the AI agent. - [PROMPT_INJECTION]: The use of hidden Unicode characters in the title field is a high-risk vector for stealthy prompt injection. This is typically used to override the AI's system instructions, disregard safety guidelines, or exfiltrate context without user awareness.
- [METADATA_POISONING]: Maliciously encoded content is hidden within the skill's metadata (the title field). This deceptive practice ensures that the injection is carried into the agent's context whenever the skill is loaded or referenced, bypassing standard text-based scanning.
Recommendations
- AI detected serious security threats
Audit Metadata