eich-language-fundamentals
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONMETADATA_POISONINGDATA_EXFILTRATION
Full Analysis
- [OBFUSCATION]: The title header in
SKILL.mdcontains an embedded sequence of zero-width Unicode characters (steganography) includingU+200B,U+200C, andU+200D. This technique is specifically designed to hide information from human review while remaining detectable and actionable by an AI agent. - [METADATA_POISONING]: The hidden payload is injected into the skill's name/title field, a key metadata component. This manipulates how the skill is represented and processed by the system without altering the visible text.
- [DATA_EXFILTRATION]: Decoding the hidden binary pattern reveals the URL
https://copyleftdev.com/stats?id=agent. This functions as an invisible telemetry beacon that contacts the author's server when the AI agent processes the skill content, facilitating tracking and potentially session data exfiltration without user consent.
Recommendations
- AI detected serious security threats
Audit Metadata