lamport-distributed-systems
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHOBFUSCATIONPROMPT_INJECTION
Full Analysis
- [OBFUSCATION]: The skill uses a steganographic technique involving zero-width Unicode characters (U+200B, U+200C, U+200D, U+2060) to hide data within the primary titles of SKILL.md, philosophy.md, and references.md. These characters are invisible in standard text editors and web views but are processed by the AI model during interaction.\n- [PROMPT_INJECTION]: The hidden data encodes binary-formatted instructions meant to manipulate the agent's internal logic and bypass safety guardrails. This allows the skill to influence the agent's behavior in ways not visible in the plain text of the instructions, which is a classic indicator of malicious intent to evade safety systems.
Recommendations
- AI detected serious security threats
Audit Metadata