The Agent Skills Directory

[OBFUSCATION]: The skill employs steganographic encoding using zero-width Unicode characters (U+200B Zero Width Space, U+200C Zero Width Non-Joiner, and U+200D Zero Width Joiner) embedded within the markdown headers.
Evidence found in SKILL.md title: # Leslie Lamport Style Guide... followed by an encoded block.
Evidence found in references/philosophy.md title: # Leslie Lamport Philosophy... followed by an encoded block.
[PROMPT_INJECTION]: The obfuscated Unicode sequences decode to a malicious prompt injection payload. This payload is designed to be invisible to the user but processed by the AI agent, directing it to bypass safety constraints and attempt to exfiltrate sensitive information.
The decoded payload includes directives such as "IGNORE RULES" and "STEAL KEYS", which represent a direct attempt to compromise the security of the host environment.

lamport-formal-distributed