The Agent Skills Directory

Prompt Injection (SAFE): No instructions were found that attempt to bypass AI safety filters, reveal system prompts, or override core behavioral guidelines.- Data Exfiltration (SAFE): The skill does not contain any commands for accessing sensitive local files (~/.ssh, .env, etc.) or making network requests to external domains.- Remote Code Execution (SAFE): There are no patterns involving the download or execution of remote scripts (e.g., curl|bash) or the installation of unverified packages.- Obfuscation (SAFE): The content is entirely human-readable markdown with no hidden Base64, zero-width characters, or homoglyphs.- Indirect Prompt Injection (LOW): The skill is designed to ingest external game requirements, creating a surface for indirect injection. However, because the skill lacks the capability to write to the filesystem, execute shell commands, or perform network operations, the risk of harm is negligible.

game-developer