The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. Ingestion points: Strings from the game server are ingested via C_GossipInfo.GetText(), C_PlayerChoice.GetPlayerChoiceOptionInfo(), and various C_TooltipInfo methods. Boundary markers: No specific delimiters or instruction warnings are present. Capability inventory: APIs are restricted to the World of Warcraft Lua environment; no system-level network or shell access is exposed. Sanitization: No sanitization logic is provided in the documentation. Findings are considered safe given the skill's primary purpose.
[DATA_EXFILTRATION]: Documented Logging and Reporting APIs. The skill references LoggingCombat and LoggingChat, which are standard WoW functions for writing logs to the local filesystem. It also covers C_ReportSystem and C_UserFeedback, which are built-in mechanisms for sending data to official Blizzard servers. These are legitimate game functions and do not constitute malicious data exfiltration.
[SAFE]: No malicious patterns, such as hardcoded credentials, obfuscation, or remote code execution from untrusted sources, were detected.

wow-api-misc-systems